The Bereavement Exclusion Debate in the DSM-5 : A History

The Premises of the Bereavement Exclusion Debate

One of the difficulties in trying to make sense of this debate is that there is a wide variation in terminology used in both the scientific literature and the public arena. For instance, what should we call bereavement-related declines in functioning that meet DSM-IV criteria for a major depressive episode but do not include the severe symptoms of suicidal ideation, psychomotor retardation, psychosis, or preoccupation with worthlessness? These presentations would be excluded according to the application of the DSM-IV bereavement exclusion rule if they occurred between 2 weeks and 8 weeks after the loss. Depending on the writer, they might be categorized as “depression,” “normal depression,” “the depression of widowhood,” “reactive depression,” “bereavement-related depression,” “uncomplicated bereavement,” “uncomplicated depressive episodes,” or “intense grief.” With so many different naming conventions, the literature on this topic very quickly becomes confusing.

To bring some uniformity to this article, by depressive episode, we mean any combination of five or more DSM depression symptoms that includes either depressed mood or diminished interest in pleasure and lasts at least 2 weeks. By uncomplicated depressive episode, we mean presentations that meet criteria for a depressive episode but would not be diagnosed if the DSM-IV bereavement exclusion rule was applied. In contrast, complicated depressive episode refers to depressive presentations that have particular severe symptoms or last longer than 2 months and would be diagnosed as a depressive disorder.

The Initiation and Development of the Professional and Public Outcry

The strategic plan for the DSM-IV revision in 1994 was to ensure that all changes are based on empirical evidence gleaned from a comprehensive review process. The leaders of the DSM-IV also believed that it was important to reduce the proliferation of diagnostic categories (Pincus, Frances, Davis, & First, 1992). They were especially cautious about introducing categories for conditions that fall below the current thresholds set for severity or duration because of the increased risk of false positives. In their view, it was also important to try to empirically evaluate possible harmful consequences of new diagnostic categories.

Allen Frances, the chair of the DSM-IV Task Force, came to believe in retrospect that they had approved changes that unintentionally resulted in harmful consequences. For instance, Frances (2013) claimed that after the DSM-IV was published, the ability of pharmaceutical companies to advertise directly to the public led to a large increase in the use of medications for attention deficit hyperactivity disorder and bipolar disorder. As argued in his book Saving Normal, this could have been prevented somewhat if they had been more concerned with tightening diagnostic categories when developing the DSM-IV.

The strategic plan for the DSM-5 was to begin a transition to dimensional models. Because dimensional models emphasize the continuity between the normal and the abnormal, they increase the attention paid to conditions that lie at the boundary of the normal and the abnormal. Boundary conditions that may represent a risk for developing a diagnosable disorder were of particular interest. For instance, there was a proposal in the DSM-5 to classify a psychosis risk syndrome. Frances (2009) opposed adding psychosis risk syndrome because he saw it as an expansion of the boundaries of psychiatric disorder to encompass a milder condition that was not an actual psychiatric disorder, thus encouraging false positive diagnoses. Eliminating the bereavement exclusion could also be seen as an expansion of diagnostic boundaries to encompass a milder condition that was not an actual psychiatric disorder.

Many of Frances’ concerns about proposals being considered for DSM-5 were published as blogs in both Psychiatric Times and Psychology Today. After the initial proposals were made public on the DSM-5 website in early 2010, Frances presented his list of the 19 worst suggestions for DSM-5 (Frances, 2010e). Eliminating the bereavement exclusion was on this list. In his view, given the inevitable overlap between grief and mild depressive disorder, eliminating the exclusion might result in general practitioners routinely diagnosing a major depressive disorder in the recently bereaved, especially if they relied only on symptom checklists and did not consider context (Frances, 2010c).

Frances’ initial posts on the bereavement exclusion initiated a debate with Zisook, and his new collaborator Ronald Pies. Pies believed that the bereavement exclusion inappropriately implied that the recent death of a loved one immunizes one from developing a depressive disorder (Pies, 2008a, 2008b, 2009). Frances (2010d) wrote that he respected the arguments of Pies and Zisook (2010b) but that medicalizing normal grief could have harmful effects. He predicted that in clinical settings many normal responses would be incorrectly classified as disordered in order to correctly diagnose the small number of people who are actually clinically depressed during bereavement. The problem, argued Frances, is that those incorrectly classified normal responses would also be subject to aggressive treatment with medication and possibly be subject to stigmatization.

In August 2010, the debate about the bereavement exclusion caught the attention of the media—beginning with a story on National Public Radio (NPR) and soon thereafter an Op-ed in the New York Times by Frances (Frances, 2010b; Speigel, 2010). Both the NPR report and the Op-ed attempted to represent both sides of the argument, but the case for keeping the exclusion was more readily framed in emotional terms.

The NPR report told the story of a woman who would visit the grave of her child and bang her head on the gravestone until her face was covered in blood. In her view, this was part of the normal process of letting her child go. This report suggested that anyone in bereavement who experienced feelings of depression for more than 2 weeks was going to be diagnosed with a psychiatric disorder and vigorously medicated, even if it was not necessary. In his Op-ed, Frances wrote that the DSM-5 proposal would substitute a “shallow Johnny-come lately medical ritual for the sacred mourning rites that have survived for millennia,” implying that the proposed change would pathologize normal grief.

By fall 2011, Allen Frances (2011b, 2011c) was writing about a user’s revolt against the DSM-5 based on letters to the American Psychiatric Association and statements of opposition authored by The British Psychological Society (2011), the Society for Humanistic Psychology (2011), and the American Counseling Association (Locke, 2011). The British letter was couched in skepticism about psychiatric diagnosis and classification in general. It emphasized the tendency in psychiatry to medicalize natural and normal reactions, but in addition to bereavement, its examples included the spectrum of depressive disorder itself. Frances (2011a) believed that this letter went too far in its critique. The other two letters were not antidiagnostic but otherwise largely echoed both Frances’ blogs and the British letter.

According to humanistic psychologist David Elkins (2011), due to the lowering of diagnostic thresholds, the DSM-5 could be dangerous to hundreds of thousands of young children, adolescents, and the elderly. The humanistic psychologists subsequently founded The Coalition to Reform DSM-5. Over 50 professional organizations of psychologists, therapists, and counselors signed on in support. They wanted an outside group of scientists and scholars not affiliated with the American Psychiatric Association to review all proposed changes to the DSM. At this point in the process, with 1 year to go before final recommendations were to be made, Frances (2011c) stated that the user’s revolt was the last and only hope for derailing the worst of the DSM-5 suggestions.

On January 24, 2012, the New York Times printed an article with the headline “Grief Could Join List of Disorders” (Carey, 2012). Five days later, an Op-ed by Gary Greenberg (2012) questioned the validity of medically classifying categories of suffering. In doing so he used the proposal to eliminate the bereavement exclusion to illustrate how the DSM labels types of suffering as diseases. That same week, on the NBC Nightly News (2012), the anchor Brian Williams reported that scientists are debating whether to classify the pain we feel after a loved one dies as a disorder and something that can be treated with medication. Similar stories also appeared in The Atlantic, Slate Magazine, The Washington Post, and The St. Louis Post Dispatch over the next few months.

For many, the most detrimental statements were published in The Lancet on February 18, 2012. The first was authored by the noted cross-cultural psychiatrist and humanist Arthur Kleinman (2012), who claimed that DSM-IV was already shockingly out of touch with the rest of the world. By diagnosing previously excluded cases with depressive disorder after only 2 months, the DSM-IV suggests that normal grief should abate in 2 months, but, said Kleinman, there is no conclusive scientific evidence that puts a time frame on the normal length of bereavement. In Kleinman’s view, the DSM-5 proposal to treat any grief as a depressive disorder represented a radical cultural framing peculiar to American academic psychiatric research. The accompanying Lancet editorial called the proposal to medicalize grief “not only dangerously simplistic, but flawed” (“Living With Grief,” 2012). An equally negative statement about medicalizing normal grief was published by Richard Friedman (2012) later that spring in the New England Journal of Medicine.

Frances (2012) reported that “the storm of opposition to DSM 5 is now focused on it’s silly and unnecessary proposal to medicalize grief.” According to Frances, “. . . it is now DSM 5 against the world.” In his interview, Kendler stated that they had unknowingly poked a hornet’s nest.

A spirited opposition to the DSM-5 proposal to eliminate the bereavement exclusion was published in a blog by Joanne Cacciatore on March 1, 2012. Inspired by earlier writings of Frances, Kleinman, and Wakefield and Horwitz, her essay was passionate—calling into question an attempt to transform a deep and existentially meaningful experience into a mental illness. To medicalize grief over the loss of a child, she suggested, is to in effect medicalize love. According to her, the proposal to eliminate the bereavement exclusion was unethical, exemplifying clinical hubris, cultural incompetence, and cultural insensitivity. For many readers, the emotional heart of the essay included five pictures of deceased children with explanations of how they died and descriptions of their parents’ reactions.

A professional grief counselor, Cacciatore added something new to the conversation by challenging the idea that suicidal thoughts are always pathological. According to her, suicidal thoughts are a normal part of mourning, especially for grieving parents.

The humanistic psychologists and Frances enthusiastically promoted her blog post. Calling it an online miracle, Frances (2012) reported that 65,000 people viewed the blog in the first 5 days after it was posted. Although, he said, the American Psychiatric Association had brushed off both the concerns of outside experts and the ridicule of the world press, perhaps a spontaneous revolt in the community of the bereaved would attract their attention.

According to DSM-5 vice chair Darrel Regier in his interview, the DSM leadership took this seriously enough that David Kupfer, the chair of the DSM-5 Task Force, agreed to talk with Cacciatore and see if her concerns could be addressed. Sid Zisook was also involved in this effort and told us that during a long cordial telephone conversation, they agreed on how devastating losing a child can be and that the pain can last a lifetime, but they were both entrenched in their positions. Cacciatore reported this came down to philosophical differences. For instance, she noted that the mental disorder label does not help in grief counseling, so she does not use the DSM. Regier reported that as Cacciatore was dismissive of the DSM and psychiatric diagnosis in general, they did not believe that further discussion with her would be productive and decided to move forward.

The Evolution of the Academic Debate

In this section, we will briefly discuss major developments in the academic debate after the proposal was made public and before the final recommendations were submitted. We do not seek to replicate literature reviews that have previously covered this material from both sides of the argument (Kendler et al., 2008; Lamb, Pies, & Zisook, 2010; Wakefield, 2013; Wakefield & First, 2012; Wakefield & Horwitz, 2016; Zisook et al., 2012; Zisook et al., 2010).

The views of Zisook and colleagues were bolstered by the addition of new studies that they believed supported their position regarding the similarity of all bereavement-related and non-bereavement-related depressive episodes (Corruble, Chouinard, Letierce, Gorwood, & Chouinard, 2009; Karam et al., 2009; Kessing, Bukh, Bock, Vinberg, & Gether, 2010).

One of the major developments in the argument for eliminating the exclusion, offered by Zisook and Shear (2009) and later by Pies and Zisook (2010a, 2010c), was the claim that intense grief and major depressive disorder are distinct and distinguishable conditions. According to them, during bereavement intense emotional distress, for example, occurs in waves and is interspersed with positive memories of the lost loved one, whereas in depressive disorder, the distress is all encompassing. In a debate with Michael First, Pies and Zisook stated that meeting five depressive symptoms every day, most of the day, for 2 weeks or more is major depressive disorder—not normal grief (First, Pies, & Zisook, 2011). In addition, people experiencing grief during bereavement usually feel that life will someday get back to normal, whereas in major depressive disorder they are typically hopeless about their future.

Early on, the different sides agreed that part of the problem was that the threshold for major depressive disorder was set too low (First et al., 2011; Frances, 2010d; Lamb et al., 2010). Lamb et al. (2010) proposed raising the diagnostic threshold by altering the duration criteria: 1 week if there was a past episode of depression; 2 weeks if symptoms include suicidal ideation, psychosis, or matched the melancholic subtype; and 1 month otherwise. This change was not implemented, likely because it would have created a large discontinuity in the research literature in the definition of major depressive disorder before and after DSM-5.

As the public controversy grew, Kendler was asked by the DSM leadership, with a very short turnaround time, to compose a statement for the DSM-5 website addressing misconceptions about the proposal to eliminate the bereavement exclusion (Kendler, 2010). Kendler argued that most people in bereavement do not develop the full syndrome of major depressive disorder, even when they have a strong reaction to the loss. He noted that the practice of excluding bereavement from a major depressive disorder diagnosis is a recent American innovation and not explicitly present in prior modern diagnostic systems. In terms of the options, extending the exclusion to all losses would be a major change and eliminating the exclusion for the loss of a loved one a small change. Finally, if a person is experiencing a depressive episode in response to bereavement, they do not have to be aggressively treated with medication. Often these bereavement-related depressions resolve on their own and watchful waiting is sometimes an appropriate clinical response.

Soon thereafter, Wakefield (2011) published a detailed response to Kendler’s DSM-5 website statement. Rather than being a recent American innovation, Wakefield claimed that for 2,400 years healers have realized that grief and depression are similar, but in grief the mood disruption is normal. Wakefield also pointed to Kendler’s recommendation that watchful waiting is often called for, noting that watchful waiting is the purpose of having a bereavement exclusion. If a change in a diagnostic criterion does not alter what mental health professionals do, there is no need to make the change.

For Wakefield, this debate was part of a larger argument about reducing false-positive diagnoses in psychiatry, with depression being a crucial example. Wakefield (2010) argued that the symptoms used to diagnose major depressive disorder were identified to distinguish depressive psychosis from schizophrenia, not to distinguish major depressive disorder from normal sadness. Lack of positive emotion and loss of energy, however, are not specific to depression; they are also common in normal sadness. If common features of normal reactions are labeled “symptoms” and then applied to the general population without considering context, many normal states will be misdiagnosed as disorders.

As the debate progressed, Ramin Mojtabai (2011) published a study that included a comparison between bereavement-related depressive episodes lasting less than 2 months to other brief depressive episodes that were unrelated to bereavement. He found that people with bereavement-related depressive episodes were less likely to experience a recurrent episode, less likely to have a comorbid anxiety disorder, and less likely to have sought treatment for depression than those with non-bereavement-related depressive episodes. In addition, people with bereavement-related brief depressive episodes were no more likely to experience a recurrent depressive episode than people with no history of depression. Mojtabai’s study was quickly followed with a replication by Wakefield and Schmitz (2013c) using another epidemiological data set that reached the same conclusion: Recurrence, historically a hallmark of mood disorder, did not occur at elevated levels in uncomplicated cases.

Both Wakefield and First believe that the publication of Mojtabai’s study should have been enough to lead the Mood Disorders Work Group to retain the bereavement exclusion in DSM-5. In their view, the finding that people with bereavement-related depressive episodes lasting less than 2 months are no more likely to experience a future depressive episode than those who have never been depressed suggests that these depressive episodes are normal states and not the manifestation of an underlying pathology.

Throughout this period and beyond, Wakefield and colleagues continued to publish new research that they believed supported retaining the bereavement exclusion (Wakefield & Schmitz, 2012, 2013a, 2013b, 2013c; Wakefield, Schmitz, & Baer, 2011). They also offered methodological critiques of the articles that Zisook and colleagues cited as supporting their position (Wakefield & First, 2012). According to First in his interview, the lack of consensus among researchers should have put a halt to the proposed change either way—the exclusion should have been neither eliminated nor extended.

The DSM Process

In the fall of 2010 at the request of Dr. Carol Bernstein, then-president of the American Psychiatric Association (APA), Kendler resigned from the Mood Disorders Work Group to chair the DSM-5 Scientific Review Committee (SRC) (Kendler, 2013). In 2009, an earlier oversight committee was created because of disquiet within the APA leadership, in part as a response to concerns reaching them from the Task Force and work groups about the how the revision process was being managed. This earlier oversight committee recommended the establishment of a more permanent committee to review the scientific evidence for any proposed changes—that is, the SRC.

The appointment to the SRC ended Kendler’s active participation in the debate. Zisook, as a consultant to the Mood Disorders Workgroup, became the primary advocate for eliminating the bereavement exclusion and met with the work group on more than one occasion. Indeed, according to Wakefield, in most instances communication about the bereavement exclusion initially went through or was referred to Zisook with potential follow-up by others, although in an attempt to bring emerging evidence to the attention of others who in his opinion might be more open-minded, Wakefield personally forwarded research supporting the bereavement exclusion to various members of the workgroup and the Task Force.

When it came time to make a final recommendation in the spring of 2012, the workgroup proposed that the bereavement exclusion should be eliminated. In his interview, workgroup member William Coryell stated that the data were relatively compelling. He supported the elimination of the bereavement exclusion because, in his words, there is no important difference between bereavement-related depressive episodes and depressive episodes related to other stressors. This view was in line with the comparison Zisook and colleagues considered important (i.e., between bereavement-related depression vs. other-stressor-related depression) rather than the comparison Wakefield and colleagues considered important (i.e., between “normal” uncomplicated episodes and “disordered” complicated episodes).

Coryell reported that Regier was worried about the controversy and wanted the workgroup to be careful, questioning them closely about their decision. Coryell believed that the outcry about labeling grief a “mental illness” was based on a misunderstanding of what was being proposed and why. Eliminating the bereavement exclusion would not result in any experience of grief being labeled a major depressive disorder.

By this point in the revision process, the DSM-5 leadership had adopted the goal of rethinking the way grief is classified. For instance, no one in the debate believed that intense grief should be over in 2 months, as the DSM-IV duration criteria were interpreted to indicate. In his interview, Regier noted that Kleinman’s editorial, in particular, called attention to this embarrassing feature of the DSM-IV. By eliminating the exclusion rule, the 2-month duration was also eliminated and with it the unintended implication that intense grief should abate after only 2 months. Another aspect of this rethinking was to better specify the nature of normal grief as distinct from grief-related major depressive disorder. Also important in this rethinking was the proposed addition of persistent complex bereavement disorder based on work spearheaded by Kathy Shear (Shear & Mulhare, 2008) and Holly Prigerson (Prigerson et al., 2009).

After the DSM-5 Task Force approved the work group’s recommendation, the next step in the process was an evaluation by the SRC in June. The committee worked entirely by phone. Kendler recused himself from participating in the SRC evaluation but was able to listen with his phone on mute. The other members concluded that the recent data about the reduced risk of recurrence for bereavement-related depressive episodes lasting less than 2 months was not enough to eclipse the extensive data about the uniformity between all stress-related depressive episodes and approved the elimination of the bereavement exclusion.

The approval of the SRC surprised several people—including Mario Maj, who in his interview stated that he did not believe that the evidence met the standards laid out in Guidelines for Making Changes to DSM-5 (Kendler, Kupfer, Narrow, Phillips, & Fawcett, 2009). According to these standards, major changes should require consistency of support across validators. In Maj’s view, even though the new research on recurrence was published late in the process, it should have been enough to prevent the deletion of the exclusion.

Maj also reported that the Mental and Behavioral Disorders chapter of the International Classification of Diseases (11th revision; ICD-11) will actually address the bereavement issue in a way that is very similar to the DSM-IV. According to its latest draft, the ICD-11 will state that “the presence of a depressive episode during a period of bereavement is suggested by persistence of constant depressive symptoms a month or more following the loss, severe depressive symptoms such as extreme beliefs of low self-worth and guilt not related to the loss of the loved one, presence of psychotic symptoms, suicidal ideation, or psychomotor retardation.”

Following the SRC, the proposal was evaluated by a second oversight committee, the Clinical and Public Health Committee (CPHC), in August. A CPHC review was triggered if there was a disagreement between the DSM-5 Task Force and the SRC or if there were concerns about the impact of a proposal on clinical practice or public health. Frances (2012) had been arguing that the proposal to eliminate the bereavement exclusion was developed by academics who were more concerned with logical consistency than clinical reality. The CPHC review was where these clinical concerns were designed to be addressed within the DSM process.

According to the information provided to us by the CPHC co-chair, Joel Yager, the committee members reviewed the arguments on both sides of the debate and were of mixed opinion on what to recommend. Despite the persuasive arguments of Wakefield and Maj, in the end the CPHC agreed to support the change because they were told that a new diagnostic criterion would be added that would extend the exclusion language by claiming that a normal response to a significant loss may resemble a depressive episode. Rather than extending the exclusion to other losses, however, the proposed new criterion would state that the presence of a depressive episode in addition to a normal response to a significant loss should be carefully considered.

Darrel Regier reported that this new addition was always intended to be an explanatory note nested under Criterion C (which states that the depressive episode is not better accounted for by the physiological effects of a substance or a general medical condition), not a separate diagnostic criterion. Ellen Frank took the lead composing the note—based on draft materials given to her by Zisook. The final version was produced using an iterative process by the work group in general, but Ellen Frank, David Goldberg, Bill Coryell, David Kupfer, and Darrel Regier spent the most time working on it.

The note is as follows:

Responses to a significant loss (e.g., bereavement, financial ruin, losses from a natural disaster, a serious medical illness or disability) may include the feelings of intense sadness, rumination about the loss, insomnia, poor appetite, and weight loss noted in Criterion A, which may resemble a depressive episode. Although such symptoms may be understandable or considered appropriate to the loss, the presence of a major depressive episode in addition to the normal response to a significant loss should be carefully considered. This decision inevitably requires the exercise of clinical judgment based on the individual’s history and the cultural norms for the expression of distress in the context of loss. (p. 134)

Regier reports that because the note required the use of clinical judgment, it was important to further amplify the clinical considerations in a footnote to provide additional guidance on how to distinguish normal grief from major depressive disorder. Also drafted by Ellen Frank and colleagues, the footnote is as follows:

In distinguishing grief from a major depressive episode (MDE), it is useful to consider that in grief the predominant affect is feelings of emptiness and loss, while in a MDE it is persistent depressed mood and the inability to anticipate happiness or pleasure. The dysphoria in grief is likely to decrease in intensity over days to weeks and occurs in waves, the so-called pangs of grief. These waves tend to be associated with thoughts or reminders of the deceased. The depressed mood of a MDE is more persistent and not tied to specific thoughts or preoccupations. The pain of grief may be accompanied by positive emotions and humor that are uncharacteristic of the pervasive unhappiness and misery characteristic of a MDE. The thought content associated with grief generally features a preoccupation with thoughts and memories of the deceased, rather than the self-critical or pessimistic ruminations seen in a MDE. In grief, self-esteem is generally preserved, whereas in a MDE feelings of worthlessness and self-loathing are common. If self-derogatory ideation is present in grief, it typically involves perceived failings vis-à-vis the deceased (e.g., not visiting frequently enough, not telling the deceased how much he or she was loved). If a bereaved individual thinks about death and dying, such thoughts are generally focused on the deceased and possibly about “joining” the deceased, whereas in a MDE such thoughts are focused on ending one’s own life because of feeling worthless, undeserving of life, or unable to cope with the pain of depression. (p. 134)

Prior to the final vote of the American Psychiatric Association Board of Trustees in early December, the various committee chairs made reports about their recommendations, and the DSM-5 leadership’s response to the outcry was explained. In the end, the bereavement exclusion was deleted.

The Influence of Mass Media

One of the more distinguishing features of this debate was the rapid transition from the academic sphere to the blogosphere, then into the domain of the media and back again. A few points stand out.

Although there was a public outcry, much of the outcry was from mental health professionals and professional organizations. Some of these organizations had a preexisting philosophical opposition in principle to psychiatric diagnosis and classification, but not all of them did. Many professionals doubtlessly became aware of the debate through newspapers, blogs, radio, and television. When these reports included interviews with recognized experts, they probably helped give the stories an air of authority and thus influenced professional opinion.

In his interview, Frances argued that advancing a proposal to eliminate the bereavement exclusion was foolish because it could readily be used to bolster the claim of groups who oppose the entire enterprise of psychiatric diagnosis and who believe that the psychiatric profession is mistakenly labelling normal people as disordered (i.e., many humanistic psychologists in the United States and advocates of the critical psychiatry movement in the United Kingdom). Frances himself had to walk a fine line in this regard to avoid being seen as colluding with opponents of psychiatry. Despite his commitment to a DSM, Frances’s opposition to the expansion of diagnostic categories to encompass milder and at-risk conditions made him a hero to those who had fundamental philosophical problems with the psychiatric diagnosis and classification.

This topic was more marketable to audiences than other controversial issues in the DSM-5 revision. Reporters generally attempted to portray both sides of the debate, but they were not interested in the details that the disputants deemed important, especially the scientific evidence that called into question the DSM’s treating bereavement-related uncomplicated depressive episodes differently than other stress-related uncomplicated episodes. In his interview, Wakefield stated that the media’s reporting did not advance the discussion.

However, sensationalism attracts an audience. Some of those interviewed believed that Wakefield and Frances fanned the flames of controversy to gain support for their position and to put pressure on the American Psychiatric Association to reverse course. The media also tended to portray those who wanted to eliminate the bereavement exclusion as being in cahoots with a corrupt pharmaceutical industry that was seeking to expand the market for psychiatric medication by labelling more people with a major depressive disorder. Both Wakefield and Frances, who were interviewed for countless stories, consistently disputed this portrayal, declaring that these changes were being proposed with good intentions and not at the behest of drug companies (Frances, 2010a; Whoriskey, 2012). In his comments on a draft of this article, Zisook noted that he had stopped using drug company money to fund his research at least 5 years before the DSM-5 process began and respected Wakefield’s and Frances’ integrity during the debate on the conflict of interest issue.

In the end, the outcry brought a lot of attention to the bereavement exclusion, but likely worked against Wakefield’s and Frances’s position due to a perception within the DSM process that the claims about psychiatrists wanting to label normal grief a mental disorder were reckless mischaracterizations. Kendler, Coryell, and Pies reported fearing that the DSM leaders would capitulate to the negative publicity and were impressed when they did not.

Was the final recommendation to eliminate the bereavement exclusion based on ignoring the hyperbole and sticking with the evidence, or was it a stubborn standing of one’s ground in response to overblown criticism? On this point, those we interviewed do not agree.

The main authoritative bodies within the DSM process—the Mood Disorders Work Group, The Task Force, SRC, and CPHC—approved the proposal to eliminate the bereavement exclusion. This consensus mattered. For something as complicated and important as a DSM, formal procedures for making decisions have been established. Had the American Psychiatric Association Board of Trustees rejected their own experts’ opinions, it would have been perceived by many as a surrendering of psychiatric and scientific authority to public relations.

One caveat regarding this conclusion is that, by and large, those advocating for deleting the bereavement exclusion had some active involvement in DSM-5 revision and those who wanted to retain the bereavement exclusion were outsiders for most of the process. Even though Kendler recused himself from his committee’s evaluation, his sentiments were clear—as was his position of authority. It is possible that had a committed opponent of the proposal played an active role in the revision process, things might have been decided differently.

The Role of Value Commitments

As a scientific debate, the question was whether uncomplicated depressive episodes following grief were valid disorders, but much of the energy of this debate involved what philosophers call nonepistemic values, especially regarding the consequences of different classificatory decisions. Would it be more harmful to misclassify normal grief as a depressive disorder or be more harmful to mislabel a depressive disorder as normal grief? These decisions are necessarily interrelated. If one follows a set of rules designed to avoid false positives, the risk of false negatives inevitably increases and vice versa. Adjudicating between these options requires making predictions about the future and balancing expected benefits and harms to both individuals and society. This is never easy.

As we noted, Wakefield and colleagues believed that false positives were more harmful, whereas Zisook and colleagues believed that false negatives were more harmful. Both groups claimed that the other side was exaggerating the potential harm. For instance, Wakefield and colleagues pointed out that excluded cases by definition do not include patients with suicidal ideation so that warning about the risk of completed suicide is scaremongering. Zisook and colleges took issue with the claim that this change would lead to an epidemic of mislabeled depression. In fact, the proposed change would only affect those experiencing an uncomplicated depressive episode during a 6-week period (from 2 weeks to 8 weeks following the loss) and who happened to present for treatment during that time period.

Another important values contrast is related to a perceived rift between the humanistic and scientific sides of modern psychiatry. As Frances said in his interview, telling someone that their grief is not a result of love but a mental disorder is degrading. For some, much of scientific psychiatry—especially biological psychiatry—is associated with an illegitimate reductionism and a lack of respect for meaningful cultural traditions. This suspicion is often reinforced by reports about extensive influence of the pharmaceutical industry on the DSM.

Interestingly, all disputants generally agreed with the humanistic perspective on the psychological importance of normal yet intense grief. They agreed that the distinction between depressive reactions following the loss of a loved one and following other severe stressful events was likely modest to nonexistent. What they did not agree on was what conclusions to draw from the empirical evidence and which evidence mattered most. They did not agree on whether it made more sense to extend the exclusion to all stress-related uncomplicated depressive episodes (Wakefield’s position) or to eliminate the exclusion criteria altogether and treat all stress-related depressive episodes as disorders (Kendler’s position).

The Nature of Psychopathology and Classification

One major point of view in this debate was that intense grief and depression are difficult to tease apart observationally because intense grief and mild depressive disorder may both lie in a fuzzy boundary region that bridges the normal and the abnormal. Allen Frances (2010b) advocated this view as seen in his claim that normal grief and major depressive disorder can be clinically completely indistinguishable. Frances’s concerns were consistent with recent thinking about what is alternatively called vertical concept creep and conceptual bracket creep, wherein the boundaries of a psychiatric disorder are expanded to encompass what were formerly considered normal conditions (Haslam, 2016; McNally, 2009).

For philosophers, fuzzy boundaries and borderline cases are potential indicators of vagueness. Vagueness is usually illustrated by the paradox of the heap, also called the sorites paradox (Keil, Keuck, & Hauswald, 2017). As introduced by the ancient Greek philosopher Eubilides of Miletus, it is evident that there is a clear distinction in kind between sand scattered on the floor and a heap of sand. If you slowly add one grain of sand to what is scattered on the floor, eventually you will create a heap. This raises a question. At what point will adding one more grain of sand transform a nonheap into a heap? Many thinkers believe that between scattered sand and a heap of sand, there is a boundary region where precise distinctions of kind cannot be made—and this is what is meant by vagueness (Zachar & McNally, 2017). According to Keuck and Frances (2017), vagueness is one possible reason why the distinction between intense grief and a depressive disorder cannot be made with precision.

As we have seen, throughout the debate there was general agreement that the presence of severe symptoms such as suicidal ideation and psychomotor retardation signaled a major depressive disorder (i.e., a psychiatric heap). Even on this point, however, people differed. For instance, Cacciatore reported that grieving parents often experience suicidal thoughts and such thoughts are not necessarily a sign of pathology. This type of complexity is consistent with vagueness.

In contrast, Zisook and Pies argued that normal grief and major depressive disorder are observationally distinct kinds. Among several other distinguishing features, in normal grief, positive and negative emotions alternate, whereas in major depressive disorder there is a more encompassing deficit of positive emotion.

Wakefield also believed that intense normal grief and major depressive disorder are at least in principle different kinds of things. For him they are different kinds because intense grief is an evolutionarily designed response and depressive disorder is a dysfunction in the person—that is, something has gone wrong with normal loss-response mechanisms. In his view, intense grief and major depressive disorder are similar in that they both feature impairment and distress (Wakefield, Schmitz, & Baer, 2010). For Wakefield, the “symptoms” of intense normal grief are not necessarily milder than in major depressive disorder. What makes them fundamentally different is the absence of an internal dysfunction in grief and its presence in major depressive disorder. Philosophically, Wakefield (2004) labels the notion that the underlying dysfunctions exist but are currently hidden as “black box essentialism.” He is seeking empirical indicators of the underlying difference he believes is there.

The problem, as noted earlier, is that internal dysfunctions can only be detected indirectly with the aid of conceptual analyses that specify when it is justified to infer that something has gone wrong. For depressive disorder, this includes an analysis of context. According to Wakefield, psychological reactions that are proportionate responses to a stressor likely signal a normal response, not a dysfunction in the person. Disproportionate responses or depressive episodes that occur in the absence of any precipitating event most likely indicate an internal dysfunction.

The problem with proportionality judgments is that they are difficult to apply reliably. Moreover, clinicians often comment that if you look closely enough, you can find potential precipitating events for every depressive episode (Lewis, 1938).

Rather than emphasizing the problem of false positives and the distinction between normal and abnormal, Kendler was interested in rebutting the view that the loss of a loved one is a unique stressor that is somehow special relative to other stressors. Kendler views depressive episodes as stable clusters of features that are sustained over time by a dappled pattern of causal processes at multiple levels of analysis. These causes exist both inside the person and external to the person (Boyd, 1991; Kendler, Zachar, & Craver, 2011). For instance, underlying genetic vulnerabilities are important causal factors, but they are not sufficient for explaining depressive episodes. Individual cases of depressive episode may be the result of different combinations of casual factors, but stress in the past 6 months is one of the more common parts of these causal packages. The loss of a loved one is an equally relevant stressor in this pluralistic model, as would be a romantic break-up or onset of a serious illness.

In contrast to Wakefield, Kendler (2016) adopts a nonessentialist position (see also Lilienfeld & Marino, 1999; Zachar, 2000; Zachar & Kendler, 2017). Rather than viewing depressive episodes as surface manifestations of a defining pathological process, Kendler views the category of major depressive disorder as a family of causally similar states that occur in varying degrees from milder to more severe. One problem with the essentialist metaphor for classification as “carving nature at the joints” is that it sets up an expectation that psychiatric classification should be more settled than is actually possible. From this perspective, Wakefield and colleagues are trying to make the distinction between intense grief and major depressive disorder more discrete than it can actually be.

Kendler also remains committed to the value of a categorical classification of psychiatric syndromes, one that emphasizes kinds, albeit kinds construed as property clusters. These property clusters can have fuzzy boundaries, but as shown by van de Leemput et al. (2014), there is often a transition to a qualitatively different state when one develops a psychiatric disorder—including a depressive disorder.

Conclusions

In this debate, the disputants agree that the suffering related to the loss of a loved one is special because it is a nearly “universal” experience had by most people at some time in their lives. They agreed that the majority of people in bereavement do not meet criteria for a major depressive episode and that the pain of grief can last a year or more, not 2 months or less. They agreed that it would be a mistake to label normal but intense distress to the loss of a loved one a mental disorder, but they did not agree on how to demarcate the normal and the disordered.

Both sides also agreed that bereavement-related uncomplicated depressive episodes that would be excluded according to DSM-IV guidelines have only minimal differences with other stress-related uncomplicated depressive episodes, for example in response to romantic breakups. Does treating these kinds of depressive episodes the same mean that both should be excluded or that both should be considered as major depressive disorders? The disputants differed strongly on which of these two changes would be best for patients and least disruptive for the continuity of research.

Looking back, it is startling how what was originally considered a technical nosologic problem, about an inconsistency in the classification of stress-related depressive episodes, resulted in such an intense controversy. Those involved expressed bewilderment at what had occurred and why. People who specialize in developing psychiatric classifications understand that such classifications are both professionally and culturally important, but rarely does power of the cultural dimension manifest like it did here. Allen Frances was likely correct that a key issue in the debate was the demarcation between normal and abnormal and the emotional power that this distinction conveys.

The mental health professions have been invested with great authority to label some kinds of suffering and impairment as disorders. Much of the power of those labels, however, lies outside of those professions. For instance, once a diagnostic construct becomes part of the culture, there is a long history of people self-applying those labels and even seeking to be diagnosed; at one time or another, neurasthenic, neurotic, addictive, and even bipolar have been sought-after labels. This is one reason why the DSM has become such an important document, although much of this impact is beyond the control of psychiatrists, psychologists, and other mental health professionals.

In contrast to past DSM revisions, the ability of the internet to transmit information and connect like-minded people together created an opportunity for them to come together and articulate what they dislike about the DSM and psychiatric nosology. The perception among some that the DSM-5 revision was more of a closed process than it had been in the past likely also played a role in fomenting opposition.

Another important issue, and one that will not go away soon, is deep disagreements among mental health professional about the consequences that adopting dimensional models have for developing diagnostic guidelines. The very nature of dimensional models calls attention to the boundary between the normal and abnormal. Those involved in revising the DSM have long been concerned about the disproportionate impact that seemingly small changes to diagnostic criteria can have, but seemingly small changes that occur in the boundary region between normal and abnormal appear to be even more contentious, both scientifically and with respect to broader values.

Those who were interviewed had diverse perceptions on the consequences of eliminating the exclusion rule. There was also clear tonal difference to these interviews. Those who believe that the bereavement exclusion should have been retained are still very passionate about it, and they do not consider the issue settled. They believe that this change was pushed through over reasonable objections. In their view, it is too soon to know about the harms created by not having a bereavement exclusion.

According to those on the other side, the dire predictions that eliminating the bereavement exclusion would artificially create an epidemic of depression and usher in an explosion of unneeded prescriptions were mistaken. In their view, this change better clarifies the difference between normal grief and a major depressive disorder and was a significant improvement.

Among the authors of this article, we have different views about how the proposal to eliminate the bereavement exclusion should have been decided. We do, however, agree that this important topic is less black-and-white than many people take it to be—including some professionals. This is, in part, because of the complex relationship between scientific evidence and diagnostic constructs but also because something as important as a psychiatric diagnostic manual tends to be a hybrid between evidence, theory, practice, policy, and values, and they may never be easy to align.

Looking ahead to the future, the American Psychiatric Association has decided to implement a continuous improvement model for DSM revisions that allows for the possibility of making changes on an ongoing basis, as long as the changes are sufficiently well supported by empirical evidence (First, Kendler, & Leibenluft, 2017). Moreover, in contrast to prior revision efforts in which changes were initiated by DSM workgroup members, it is expected that most proposals for change will be submitted by clinicians, researchers, and organizations that are external to the DSM process. It should be noted this continuous revision process was not established to allow for those dissatisfied by DSM-5 decisions to simply “relitigate” those decisions in front of a different group of judges. Instead, submissions that aim to reverse a DSM decision must present new and strong empirical data that were lacking the first time around.