DSM-5: proposed changes to depressive disorders

Abstract

The Diagnostic and Statistical Manual of Mental Disorders (DSM) is currently undergoing a revision that will lead to a fifth edition in 2013. Proposed changes for DSM-5 include the creation of several new categories of depressive disorder. Some nosologists have expressed concern that the proposed changes could yield many ‘false-positive diagnoses’ in which normal distress is mislabeled as a mental disorder. Such confusion of normal distress and mental disorder undermines the interpretability of clinical trials and etiological research, causes inefficient allocation of resources, and incurs risks of unnecessary treatment. To evaluate these concerns, I critically examine five proposed DSM-5 expansions in the scope of depressive and grief disorders: (1) a new mixed anxiety/depression category; (2) a new premenstrual dysphoric disorder category; (3) elimination of the major depression bereavement exclusion; (4) elimination of the adjustment disorder bereavement exclusion, thus allowing the diagnosis of subsyndromal depressive symptoms during bereavement as adjustment disorders; and (5) a new category of adjustment disorder related to bereavement for diagnosing pathological non-depressive grief. I examine each proposal’s face validity and conceptual coherence as well as empirical support where relevant, with special attention to potential implications for false-positive diagnoses. I conclude that mixed anxiety/depression and premenstrual dysphoric disorder are needed categories, but are too broadly drawn and will yield substantial false positives; that the elimination of the bereavement exclusion is not supported by the evidence; and that the proposed elimination of the adjustment-disorder bereavement exclusion, as well as the new category of grief-related adjustment disorder, are inconsistent with recent grief research, which suggests that these proposals would massively pathologize normal grief responses.

Key words::

• Adjustment disorder
• Bereavement
• Diagnosis
• DSM-5
• False positives
• Grief
• Harmful dysfunction
• Major depression
• Mixed anxiety/depression
• Premenstrual dysphoric disorder
• Validity

Introduction

The American Psychiatric Association’s Diagnostic and Statistical Manual of Mental Disorders (DSM)1 is currently undergoing a revision that will yield the fifth edition (DSM-5) in 2013. The DSM-5 Mood Disorders Work Group has proposed many changes to depressive disorder categories2, which are likely to substantially impact clinical practice. Of particular interest are, expansions of the domain of disorder to encompass conditions previously classified as normal variation, which provide new targets for pharmacotherapy and other treatment. They also present the clinician with novel challenges in assessment and treatment planning, especially regarding medication advisability for cases on the normal/disorder boundary. Of particular concern are potential false-positive diagnoses of normal distress as disorder, leading to unnecessary or overly aggressive treatment. The false-positives concern has been prominent in recent debates about DSM-53–5, yet the DSM-5 website and literature reviews do not extensively address this issue.

Here I review some of the DSM-5 proposals for new categories of depressive and grief disorders or expansions of existing categories. Depressive feelings are common in the community, so even minor changes in diagnostic thresholds can substantially increase the number of diagnosable individuals. Such threshold changes deserve special scrutiny given recent diagnostic expansionism. Prior to the DSM-III’s revolutionary symptom-based descriptive diagnostic criteria6, clinicians were commonly taught that depressive disorder – or melancholia – is a relatively rare and severe occurrence, afflicting perhaps 2–3% of the population over a lifetime7. In contrast, recent longitudinal community studies suggest that well more than half the population satisfies current DSM-defined major depression criteria at some point8–10 – and this estimate ignores subsyndromal, dysthymic, and other related conditions. These findings raise questions about whether diagnostic thresholds are too low. Suggestions have been put forward for better discriminating normal distress or mild conditions from more seriously pathological depressive syndromes clearly warranting treatment11–16, but DSM-5 proposals move in the opposite direction of encompassing more conditions as pathological.

I focus here on five proposals expanding depressive and grief disorders, examining them for their face validity in distinguishing between disorder and normality: (1) a new mixed anxiety/depression disorder category; (2) a new premenstrual dysphoric disorder category; (3) an expansion of major depression by eliminating the bereavement exclusion; (4) an expansion of adjustment disorder (AD) by eliminating the AD bereavement exclusion and thus allowing subsyndromal depressive symptoms during bereavement to count as a disorder; and (5) a new bereavement-related adjustment disorder category proposed by the Anxiety Disorders Work Group.

Note that the DSM-5 revision is still in process, so these proposals might not end up appearing in DSM-5 in their current form. Also, many other important proposals are not addressed here. These include various changes to bipolar disorders as well as the new ‘disruptive mood dysregulation’ category aimed at stemming overdiagnosis of temper-tantrum prone children as bipolar (but having its own potential for false-positive diagnoses of children17). Additionally, there are various proposed changes aimed at increasing clinical utility by capturing additional information but not affecting primary diagnosis, such as a mixed-features specifier for major depressive episodes that include subthreshold manic symptoms, a symptom severity scale, and an anxiety comorbidity dimension.

The identification of false positives in psychiatry is challenging and potentially more controversial than in most areas of physical medicine, where laboratory tests can demonstrate a spurious versus valid diagnosis. The DSM’s definition of mental disorder requires that, to be a disorder, symptoms must be caused by a dysfunction in the individual, rather than being a matter of deviance, social undesirability, or expectable reactions to stress. This requirement provides the primary guidance in judging diagnostic validity. When dysfunction is absent, symptoms represent normal-range variants of distress or eccentricity and should be classified as DSM ‘V Code’ conditions, that is, non-disorders for which clinicians are sometimes consulted, such as distress due to occupational or family conflicts, routine delinquency, normal sadness, and illiteracy. When a condition that is not due to a dysfunction is nonetheless misdiagnosed as a disorder, that is a ‘false-positive’ diagnosis. Determining whether something has gone wrong with internal mental functioning is something clinicians do every day, based on implicit assumptions about the proper functioning of psychological and biological processes. However, because much presently remains unknown about such processes, judgments about dysfunction, including those put forward in this analysis, must typically rely on face validity plus diverse sources of circumstantial evidence. Despite these challenges, the avoidance of false-positive diagnoses is crucial not only clinically in terms of prognosis, treatment planning, and informed consent, but also regarding research into etiology and treatment outcome. Mixing together normal and disordered conditions renders the meaning of research results difficult to interpret or generalize.

In focusing my commentary on false positives, I set aside the equally important challenge of limiting false-negative diagnoses, in which mental disorder is mistakenly considered normal distress. False negatives raise different kinds of issues and they generally concern a different set of individuals than those falling under concerns about false positives.

Mixed anxiety/depression

Along with sadness, anxiety has traditionally been considered a core symptom of depression. For example, Hippocrates defined melancholia as ‘fear and sadness lasting a long time’18, British vicar Robert Burton defined melancholia in his classic Anatomy of Melancholy as ‘fear and sadness, without any apparent occasion’19, and Emil Kraepelin specifies ‘anxious tension’ as a typical symptom of depressive illness20.

However, starting with DSM-III, the attempt to sharply distinguish depressive and anxiety syndromes led to elimination of anxiety as a symptom of major depression. Subsequently, psychiatry ‘discovered’ a mysteriously high comorbidity between depression and anxiety, and grappled with how to address this problem. The issue became more urgent as studies showed that depression with comorbid anxiety has increased risk of various negative outcomes. The DSM-5’s proposed mixed anxiety/depression category reunites depressive and anxiety symptoms within a new depressive category.

Conditions not satisfying full syndromal criteria for either depressive or anxiety disorders but containing both types of symptoms currently must be classified as subsyndromal disorders falling under mood disorder not otherwise specified (NOS) or anxiety disorder NOS or both. The patient is implausibly seen as having two distinct subsyndromal disorders rather than one syndromal disorder. Thus, in principle, a mixed depression/anxiety disorder is a welcome addition to the manual.

The DSM-5 proposal builds on earlier criteria proposed for further study in DSM-IV’s appendix that require 1 month of dysphoric mood and four out of ten specified symptoms. However, the presumed anxiety-symptom criteria that the appendix’s proposal added to major-depression symptoms are in fact conceptually ambiguous between depressive and anxiety symptoms (e.g., worry, easily crying, negative anticipatory bias, irritability) and so do not face validly indicate mixed anxiety/depression as opposed to standard major depression.

The DSM-5 proposal attempts to rectify this situation by specifying that the patient must have three or four symptoms of major depression (including either depressed mood or anhedonia) and anxious distress. Anxious distress is defined as ‘having two or more of the following symptoms: irrational worry, preoccupation with unpleasant worries, having trouble relaxing, motor tension, fear that something awful may happen’21.

The problem is that, because the proposed DSM-5 criteria are quite different from those in DSM-IV’s appendix, there has been virtually no research on them. Lacking any evidential anchoring, the proposal’s face validity becomes crucial.

The proposed criteria require not the usual five out of nine depression symptoms but rather three out of nine depression symptoms plus two out of five anxiety symptoms (for only 2 weeks, reduced from the DSM-IV appendix’s 1 month). This offers a much enlarged universe of symptoms from which criteria may be met. This would not be problematic if most of the symptoms were pathognomonic for disorder. However, unlike, say, psychomotor retardation, morbid preoccupation with worthlessness, or suicidal ideation, which inherently suggest pathology, the three depressive symptoms might be general distress indicators such as fatigue, insomnia, and trouble concentrating on usual tasks. The anxiety criteria suffer from the same problem; under negative circumstances, ‘preoccupation with unpleasant worries,’ ‘trouble relaxing,’ and ‘fear that something awful may happen’ can be normal and even adaptive responses.

Anxiety disorder diagnostic criteria are sometimes formulated to avoid false positives by requiring that the anxiety must be ‘excessive’ or ‘unreasonable’ or, as proposed for DSM-5 phobias, ‘out of proportion to the actual danger’22. However, except for the symptom of ‘irrational worry,’ the proposed anxiety symptoms have no such qualifiers. They are thus open to false-positive diagnoses of individuals experiencing normal distress in response to real threats. Such qualifiers might be added to improve validity.

Premenstrual dysphoric disorder

Proposed criteria for premenstrual dysphoric disorder specify a diagnostic threshold of five out of 11 listed symptoms that occur during the week before menses and subside in the week following menses, as follows:

DSM-5 proposed diagnostic criteria for premenstrual dysphoric disorder

In most menstrual cycles during the past year, five (or more) of the following symptoms occurred during the final week before the onset of menses, started to improve within a few days after the onset of menses, and were minimal or absent in the week postmenses, with at least one of the symptoms being either (1), (2), (3), or (4):

1. Marked affective liability (e.g., mood swings; feeling suddenly sad or tearful or increased sensitivity to rejection)

2. Marked irritability or anger or increased interpersonal conflicts

3. Markedly depressed mood, feelings of hopelessness, or self-deprecating thoughts

4. Marked anxiety, tension, feelings of being ‘keyed up’ or ‘on edge’

5. Decreased interest in usual activities (e.g., work, school, friends, hobbies)

6. Subjective sense of difficulty in concentration

7. Lethargy, easy fatigability, or marked lack of energy

8. Marked change in appetite, overeating, or specific food cravings

9. Hypersomnia or insomnia

10. A subjective sense of being overwhelmed or out of control

11. Other physical symptoms such as breast tenderness or swelling, joint or muscle pain, a sensation of ‘bloating,’ weight gain.

In addition, the criteria require that the symptoms be confirmed over at least two cycles, that symptoms occur when the woman is not taking oral contraceptives, that the symptoms not be due to other mental disorders or to substance use or a general medical condition, and that the symptoms are associated with ‘clinically significant distress or interference with work, school, usual social activities or relationships with others (e.g., avoidance of social activities, decreased productivity and efficiency at work, school or home).’ The latter ‘clinical significance’ criterion, designed to prevent false positives, has the obvious weakness that normal discomfort during perimenses often involves distress and interference with or avoidance of some usual activities.

This category is controversial due to concerns about pathologizing normal female biology. However, any biological process can go wrong. Dysphoria preceding menstruation may be profound enough to suggest pathology, so an appropriate disorder category could be clinically useful, especially given the existence of potentially helpful medications.

However, the many ways to symptomatically qualify for diagnosis under the proposed criteria include some that seem of questionable validity in distinguishing normal variation in the symptoms of perimenses from disorder. For example, a woman qualifies for disorder if in the days preceding menses she generally experiences marked emotional lability (e.g., increased sensitivity to rejection), marked feelings of being ‘on edge’; easy fatigability; specific food cravings; and breast tenderness, which seems potentially within normal range. Qualifying symptoms also include, for example, self-deprecating thoughts, decreased interest in usual activities, and insomnia, all of which could easily occur normally in those feeling edgy, fatigued, and physically uncomfortable. Nor is it clear that including typical premenstrual physical symptoms as a criterion increases specificity in distinguishing normal from psychiatrically disordered perimenses.

Another problem is that, in the context of the normal symptom variations in perimenses, diagnostic thresholds may be invalidly low. For example, the criterion ‘increased interpersonal conflicts’ could be due simply to normal-range increased irritability or sensitivity in a context of ongoing relationship difficulties. This criterion could be strengthened by requiring conflict that emerges during the episode and is not just an exacerbation of ongoing conflict. Similarly, ‘self-deprecating thoughts’ could easily encompass negative thoughts women often have due to transient weight gain and other physical changes during perimenses, and validity could be strengthened by requiring generalized deprecation or deprecation markedly disproportionate to circumstances.

The Work Group acknowledges the concern that ‘a disorder that focuses on the perimenstrual phase of the menstrual cycle may ‘pathologize’ normal reproductive functioning in women’24. They answer that the diagnosis exonerates most women from being pathologically impaired because most are not classified as disordered: ‘We felt that the prevalence statistics clearly indicate that PMDD is a condition that occurs in a minority of women. As such, it would be inappropriate to generalize any disability to women in general. In fact, a DSM diagnostic category for women who experience marked symptoms and impairment perimenstrually highlights the fact that most women do not experience such symptoms’24.

This rationale for pathologizing a large minority of women based on several days of distressing symptoms during the menstrual cycle is not scientifically compelling, to say the least. After all, any diagnostic criteria, no matter how invalid, define those who fall outside the criteria as normal. The question is whether substantial numbers of individuals satisfy the criteria but are in fact within normal range, thus are being mistakenly classified as disordered. The Work Group’s rationale does not address this question, and the criteria suggest a substantial false-positives problem.

Elimination of the major depression bereavement exclusion

The severity of normal grief after losing a loved one has long been recognized and often depicted in literature across cultures, from Gilgamesh and The Iliad to recent works by Roland Barthes, Joan Didion and Joyce Carol Oates25–27. Prospective studies have established that depressive symptoms are commonly part of such normal reactions to loss of a loved one, and that bereaved individuals frequently reach symptom levels satisfying major depression diagnostic criteria within the months after loss, but such reactions are generally transient28. Moreover, although recently bereaved individuals frequently consult physicians for help with symptoms such as insomnia, they generally do not think of themselves as mentally disordered when experiencing symptoms such as sadness, trouble sleeping, fatigue, diminished interest in activities, and trouble concentrating on usual tasks28. On the other hand, public campaigns to increase awareness of depressive disorder as well as direct-to-consumer advertising regarding antidepressant medication have increased public concern about whether emotions following loss represent a dangerous mental disorder, and patients do often seek reassurance on this point by consulting a physician.

Concern to avoid mistaking normal grief for psychiatric disorder in patients coming in for consultation after loss led to the incorporation in DSM-III and subsequent editions of the ‘bereavement exclusion’ in criteria for major depression (Criterion E of major depressive episode), which excludes from diagnosis certain bereavement-related depressive episodes that are ‘uncomplicated’ in the sense that they are fairly brief and do not contain certain severe symptoms. However, it is well-established that bereavement can precipitate true depressive disorders, so not all such episodes can be excluded. To minimize false-negative diagnoses, the bereavement exclusion is formulated to distinguish those post-loss depressions that are likely normal from those that are likely disorders. It specifies that a bereavement-related depressive episode will not be excluded and will still be diagnosed as major depression if it lasts longer than 2 months or if it contains any one of several pathology-suggestive symptoms, including suicidal ideation, psychomotor retardation, and sense of worthlessness.

The DSM-5 Mood Disorders Work Group proposes to eliminate the exclusion entirely, thus newly qualifying millions of bereaved individuals each year who have depressive symptoms for just 2 weeks after loss of a loved one for depressive disorder diagnosis. The only rationale offered for this change is that ‘evidence does not support separation of loss of loved one from other stressors’29. That is, bereavement-related depressive episodes are similar to other depressive episodes, so it is illegitimate to distinguish bereavement-related episodes in an exclusion. This argument incorrectly presupposes that an emotional reaction can't be normal in response to one set of circumstances but pathological when experienced in other circumstances. Only one literature review is cited in support of the similarity claim30. That review examined all bereavement-related depression, not the very limited set of conditions excluded by the bereavement exclusion. Thus, the cited review does not address the crucial question of whether the exclusion’s distinction between those transient, mild episodes that should be excluded versus those that are true disorders is valid.

The proposal to eliminate the bereavement exclusion triggered a debate over the exclusion’s validity that has generated many reviews and editorials supporting the proposal generated by overlapping authors31–35. However, these reviews have generally relied on repeated citations of a few research studies that are of questionable relevance. Most of the cited research does not sample DSM-defined bereavement-excluded cases36–38; another much-cited prospective study’s sample size of n = 5 bereavement-excluded individuals is acknowledged by the authors to be too small for scientific generalization39.

Two frequently cited epidemiological studies did examine large samples of DSM-defined excluded cases40–41. The central finding of both studies was that, if one applied identical exclusion criteria for ‘uncomplicated’ cases to both bereavement-related cases and other-triggered depression cases, such as depressive feelings triggered by job loss or relationship disruption, then excluded bereavement-related depressions are quite similar on a range of validators to other-triggered depressive episodes satisfying the same exclusion criteria.

However, the meaning of this ‘parity’ result was ambiguous. One side argued that, since uncomplicated bereavement-related depression is not currently considered a disorder, the results imply that uncomplicated reactions to other stressors are likely normal and should be excluded from major depression diagnosis as well. The other side argued that, since uncomplicated reactions to other stressors are currently considered disorders, the parity result proves that the DSM has been wrong all along in excluding uncomplicated grief-related depression, which should be considered a disorder as well. This dispute ended in an impasse. One cannot persuasively use what the DSM does or does not now consider a disorder to argue one way or the other about the validity of the DSM bereavement exclusion, because that approach begs the question at issue of the proper normal/disorder boundary. A defensible conclusion must be based on independent evidence using persuasive validators, not on diagnostic tradition.

A recent new wave of research focuses on the most persuasive validator, recurrence, and offers a surprisingly definitive answer to the conundrum posed by earlier studies. The recurrent course of major depression is generally considered one of its most distinctive validators. Two recent prospective epidemiological analyses provide unambiguous support for the hypothesis that excluded bereavement-related depressive episodes are different from other depressive episodes and best considered normal variants42,43. Both studies classify types of depressive episode at time 1 and then examine the rates of recurrence at time 2. The two studies use different data sets, different follow-up time periods (1 year versus 3 years), and somewhat different analytic methods, but come to the same conclusion. First, as expected, those who had lifetime depression at time 1 have a dramatically higher recurrence rate at time 2 than the occurrence rate at time 2 among those who have no history of depression at time 1. Second, there is a difference between excluded episodes and other depression; those with only excluded episodes at time 1 have substantially and significantly lower recurrence rates at time 2 than other types of depressive episodes. However, the crucial finding in both studies is a third finding, that among those who had lifetime excluded bereavement-related depression at time 1, the rate of depression recurrence at time 2 is not significantly greater than the rate of occurrence of depression at time 2 in the general population reporting no history of depressive disorder at time 1. The discovery and replication of the finding that excluded bereavement-related depression cases do not recur at over the general population rate of occurrence is a strong empirical argument for considering them normal-range emotional reactions rather than depressive disorders.

Jan Fawcett, the Chair of the DSM-5 Mood Disorders Work Group, credits treatment response as the sole reason for eliminating the exclusion44. This claim is based on a single study in which 22 bereaved individuals satisfying DSM-IV MDD criteria about 2 months post-loss were treated with buproprion-SR for 2 months; 13 subjects experienced a reduction of ≥50% on Hamilton Depression Rating Scale scores45. Given the small sample size and the fact that the study contained no control group yet examines a condition with notoriously high placebo response rates during a bereavement period when prospective studies show precipitous drops in symptoms, the modest ‘response rate’ is impossible to interpret. For comparison purposes, in Clayton’s prospective study of normal bereavement, in the period from the first 2 weeks to on average about 3 months after loss, the six highest-prevalence depressive symptoms decreased as follows: depressed mood, 87% to 12%; sleep disturbance, 85% to 27%; crying, 79% to 12%; difficulty concentrating, 47% to 27%; loss of interest in TV, news, friends, 42% to 19%; anorexia and/or weight loss, 49% to 27%28. Thus, the reported outcomes in the uncontrolled treatment study cited by Fawcett are consistent with trajectories of resolution for normal grief symptoms without medication.

Moreover, even if it should turn out that bereavement-related depressive symptoms respond to medication, the inference to pathology remains uncertain given that many normal conditions respond to medication. Anxiolytics, stimulants, hypnotics, and many other psychoactive medications are known – and regularly used – to influence normal as well as disordered function, and social efforts have been made to limit their excessive use to manipulate normal variations in functioning. The same may be true of antidepressant medication. Studies offer divergent findings as to whether normal mood is made more positive by SSRIs46. However, there is suggestive evidence that normal negative affect may be influenced by antidepressant medication47. This will depend on the precise pathway of action of specific medications; brain activity during depressive disorder overlaps in many respects with normal sadness48, so presumably some pathways will produce changes in both. Diminished diagnostic differentiation in DSM-5 between bereavement-related depressive feelings and major depression will require increased physician judgment in deciding when intervention is appropriate to address disorder or reduce normal suffering.

Those arguing for elimination of the bereavement exclusion have raised the specter of suicide in excluded bereavement-related depressions49–50, in one case going so far as to declare to a general audience on public radio that ‘I’d rather make the mistake of calling someone depressed who may not be depressed, than missing the diagnosis of depression, not treating it, and having that person kill themselves’51. Such claims fail to take into account that suicidal ideation as well as sense of worthlessness must be ruled out for a bereavement-related episode to qualify for exclusion, precluding usual risk factors for suicide attempt in bereavement. The existing evidence provides no basis for suggesting an elevated suicide rate in bereavement-excluded depression.

Eliminating the adjustment disorder bereavement exclusion, thereby allowing subsyndromal depressive symptoms during bereavement to be diagnosed as adjustment disorder with depressed mood

DSM adjustment disorder (AD) diagnostic criteria contain no duration or symptom-number thresholds, and one of the symptom specifiers for AD is ‘with depressed mood’ (‘when the predominant manifestations are symptoms such as depressed mood, tearfulness, or feelings of hopelessness’). Thus, following a stressor, even transient and minimal depressive symptoms can be diagnosed as an adjustment disorder. However, subsyndromal depressive symptoms such as sadness and insomnia are extremely common during normal grief. For example, in Clayton’s classic prospective study of normal grief, in the first weeks post-loss, 87% of bereaved subjects reported depressed mood, 85% sleep disturbance, 79% tearfulness, and about half reported each of diminished interest in usual activities, difficulty concentrating, and lessened appetite28. Consequently, to avoid massively pathologizing grief, the DSM Adjustment Disorder criteria currently include a bereavement exclusion, ‘the symptoms do not represent bereavement.’ Without the bereavement exclusion, virtually the entirety of Clayton’s normal-grief sample would satisfy AD-with-depressive-mood symptom criteria.

Nonetheless, the DSM-5 Anxiety Disorders Work Group, which has responsibility for the adjustment disorder category, proposes eliminating the AD bereavement exclusion, thus allowing the diagnosis of subsyndromal depressive symptoms during grief as an AD. The rationale for this change is obscure to say the least. Strain et al.’s review for the Work Group admits that ‘there is no research that has specifically investigated the bereavement exclusion in AD’, yet defends the proposal with the novel argument that there is ‘no evidence to the contrary’52. The DSM-5 website’s ‘rationale’ section offers the explanation that ‘elimination of the bereavement exclusion with respect to mood disorders, has led to the decision to also eliminate this exclusion for adjustment disorders’53. But why should eliminating the major depression bereavement exclusion – which at least concerns a recognized disorder – justify reconceptualizing subsyndromal depressive grief symptoms, traditionally considered consistent with normal sadness, as adjustment disorders?

The website explains that it is likely that grief-related subsyndromal depressive feelings resemble subsyndromal depressive feelings after other stressors, like job loss or relationship disruption, which currently qualify as AD. However, whether such reactions to other stressors when transient and mild are themselves always disorders is questionable and has never been seriously addressed in the DSM revision process. It is widely recognized that current AD criteria tend to straddle the normal/disorder boundary54. Moreover, using sheer symptom resemblance taken out of context of an understanding of normal human emotional responses to loss makes no sense as a validation procedure; it is as absurd as labeling intense anxiety when one’s life is immediately threatened as a disorder just because the anxiety resembles the anxiety of agoraphobics or panic-disordered individuals55. Depressive feelings and anxiety are biologically designed responses that in some contexts and in some intensities are normal. The website’s argument reifies symptoms into disorders, begging the question of the nature of normal human nature and pathologizing grief in the process.

Proposed DSM-5 AD criteria do also require that the symptoms be clinically significant, as ‘evidenced by either marked distress in excess of what would be proportionate to the stressor (taking account of context and cultural setting) or significant impairment in social, occupational, or other important areas of functioning.’ However, grief is inherently intensely distressing, and the emotional intensity of acute grief routinely carries with it some role impairment. The clinical significance requirement will not block the tidal wave of false-positive diagnoses potentially created by the proposed elimination of the AD bereavement exclusion.

Adjustment disorder related to bereavement

Typical non-depressive symptoms in response to loss of a loved one, such as yearning, disbelief, and feelings of emptiness, are currently excluded as symptoms of adjustment disorder (AD) by the AD bereavement exclusion due to their presumed normality. Having eliminated the AD bereavement exclusion, the DSM-5 Anxiety Disorders Work Group proposes adding a new AD symptom specifier, ‘related to bereavement,’ to allow such non-depressive grief symptoms after loss to constitute adjustment disorders. In addition to standard adjustment-disorder requirements (symptoms must start within 3 months of the loss and cause either excessive distress or role impairment), adjustment disorder related to bereavement requires that ‘for at least 12 months following the death of a close relative or friend, the individual experiences on more days than not intense yearning/longing for the deceased, intense sorrow and emotional pain, or preoccupation with the deceased or the circumstances of the death. The person may also display difficulty accepting the death, intense anger over the loss, a diminished sense of self, a feeling that life is empty, or difficulty planning for the future or engaging in activities or relationships. Mourning shows substantial cultural variation; the bereavement reaction must be out of proportion or inconsistent with cultural or religious norms’53.

These criteria appear to require only one symptom, such as yearning, for diagnosis, given that the initial symptom criteria are disjunctively linked by ‘or’ and the rest of the symptoms ‘may be’ present. Thus, anyone who most days still experiences intense yearning for the lost person 1 year post-loss is diagnosable with a mental disorder. But whether requiring one symptom or a few, these criteria raise serious concerns about validity.

Grief researchers have carefully studied the psychometric properties and validity of many possible criteria sets for pathological grief over the years56, and offered a variety of proposals57. The proposed DSM-5 criteria are weaker than any previously studied or proposed criteria set. Consequently, there is no research base for the specific DSM-5 proposal. (The Work Group does also propose placing in the DSM-5's appendix for future study a stronger and more restrictive criteria set for a new bereavement-related disorder, closer to the criteria sets proposed by grief researchers. However, because this would not constitute a change in Axis 1 or 2 disorders, the proposed criteria are not considered here.)

The grief research on more stringent criteria sets does offer some grounds for estimating how commonly the proposed DSM-5 criteria would be satisfied by the bereaved. The research suggests that symptoms of grief commonly continue well beyond 12 months. Thompson et al. conclude that ‘the experience of grief persists for at least 30 months in both older men and women who have lost their spouse’58. Prigerson et al.’s data indicate that the average yearning frequency at 1 year post-loss among all bereaved is about every other day, which is not that distant from the DSM’s proposed threshold for pathology of having yearning ‘more days than not,’ suggesting that under the proposed DSM-5 criteria substantial percentages of normal grievers would satisfy the criteria and qualify for diagnosis59. Horowitz et al. required at least three symptoms at 14 months post-loss for their proposed diagnosis, which is much more demanding than the proposed DSM criteria, and found that 41% of their bereaved sample qualified for diagnosis60; all of these 41% and many more besides would likely qualify as disordered under the weaker DSM-5 proposal. The false-positives potential in the DSM-5 proposal is thus very large indeed, probably encompassing well more than half of all bereaved – and these potentially diagnosable individuals are the bereaved who are not necessarily depressed.

The DSM-5 proposal seems out of step with contemporary research on grief, which suggests that the grief process is not, as classically portrayed, a step-wise process that leads to resolution of pain within a specified period, but rather an individually constructed compromise in which pain never fully resolves but is compartmentalized in order to move on with one’s life61. In arbitrarily legislating the duration of normal grief, the proposal also ignores the prescient warning of the preeminent grief researcher, John Bowlby, who, while amply recognizing the existence of pathological grief reactions, cautioned against precisely the sort of premature pathologization of grief that would result from the DSM proposal. Bowlby believed that widows who come close to fully recovering their pre-loss functioning ‘are more likely to take 2 or 3 years to do so than a mere 1’ and that ‘clinicians sometimes have unrealistic expectations of the speed and completeness with which someone can be expected to get over a major bereavement’. He chided clinicians for rushing to pathologize the resulting distress: ‘[T]here is a tendency to under-estimate how intensely distressing and disabling loss usually is and for how long the distress, and often the disablement, commonly lasts. Conversely, there is a tendency to suppose that a normal healthy person can and should get over bereavement not only fairly rapidly but also completely’62. The DSM-5 Task Force would do well to review its proposed adjustment-disorder-related-to-bereavement diagnostic criteria with Bowlby’s observations on validity firmly in mind.

Conclusion

The proposed DSM-5 expansions of depressive and grief disorders do not adequately address issues concerning the boundary between disorder and normality. Consequently, the DSM-5 proposals pose threats of substantial false-positive diagnoses. While it is beyond the scope of this editorial to suggest new studies or field trials to address the issues raised here, the results of the above analysis combined with the level of concern about false positives expressed in the literature suggests that field trials and supplemental studies are necessary to establish the impact of proposed changes on the diagnostic status of mild, transient, and face-validly normal conditions that are currently not pathologized.

DSM diagnostic criteria are commonly applied in checklist screening instruments in physicians’ offices and in schools, in community prevalence studies that guide policy and funding priorities, and in forensic settings. In none of these contexts is clinical judgment always available to correct for flaws in the criteria; the validity of the criteria really matters. Moreover, DSM criteria are used for sample selection in clinical trials. In order to proceed efficiently, samples for such trails tend to be chosen toward the lower end of severity, close to the boundary between disorder and normality. Flawed criteria subject to substantial false positives may lead to lack of power of trials to demonstrate the effectiveness of medications for disorders due to high placebo and spontaneous remission rates of those with normal distress. If samples are mixes of normal and disordered individuals, results of trials may be difficult or impossible to interpret in terms of their generalizability to truly disordered populations. The same concerns about heterogeneous samples apply even more to research on causation of mental disorder that supports longer term efforts to develop treatments.

Invalidities in diagnostic criteria also present clinicians with especially challenging assessment and treatment dilemmas. If the diagnostic criteria do not successfully distinguish the disordered from the normal distressed, yet establishing prognosis and making responsible medication decisions with the informed consent of the patient to some extent depend on making such distinctions, then the clinician will be compelled to accept responsibility for second-guessing the criteria and not diagnosing a condition that meets criteria. Empirically and conceptually valid criteria provide the most clinically useful basis for treatment decisions and should be the focus of further efforts to refine DSM-5 proposals.

Transparency

Declaration of funding

This editorial was not funded.

Declaration of financial/other relationships

J.W. has no relevant financial relationships to disclose. CMRO peer reviewers on this manuscript have disclosed that they have no relevant financial relationships.