Opening Pandora's box: how DSM-5 is coming to grief


In mourning, it is the world which has become poor and empty; in depression it is the ego itself.

– Freud [1].

Until recently, DSM-5 architects had sought to reposition grief within the depressive disorders: a proposal generating multiple reviews and critiques [2-4], and evoking concerns about a risk of ‘pathologising’ bereavement. Friedman [4] captured a second common ‘medicalization’ concern, whereby a diagnosis of ‘major depression’ could be made in those experiencing normal bereavement after only 2 weeks of mild depressive symptoms, and which could then lead to ‘unnecessary treatment with antidepressants and antipsychotics’.

This discussion paper considers historical changes in categorizing grief within recent DSM manuals, focuses on broad parameters that differentiate grief and depression, and then considers implications to a DSM-5 website comment written by Professor Kendler [5], before I offer a revisionist position.

DSM-IV vs. DSM-5 criteria

DSM-IV criteria for a major depressive episode are commonly viewed as excluding grief states, however this is not exactly the case. The relevant criterion (E) states that ‘The symptoms are not better accounted for by Bereavement,’ unless ‘the symptoms persist for longer than 2 months or are characterized by marked functional impairment, morbid preoccupations with worthlessness, suicidal ideation, psychotic symptoms, or psychomotor retardation after the loss’. In essence, a DSM-IV diagnosis of Major Depression can be made following bereavement on the basis of (i) persistence (exceeding 2 months), (ii) severity (for instance suicidal ideation) and (iii) psychotic or melancholic depressive features.

The DSM-5 Mood Work Group initially sought to remove the formal DSM-IV bereavement criterion on the basis [6] that ‘evidence does not support separation of loss of loved ones from other stressors.’ But such evidence was generated from only one review, and which was subsequently criticized by Wakefield [3] on the basis that it examined ‘all bereavement-related depression, not the very limited set of conditions excluded by the bereavement exclusion.’

Following such criticisms, the DSM-5 committee regrouped and, still evidencing their commitment to reshape the perceived anomaly, drafted a provisional April 2, 2012 DSM-5 website note [7] seeking to guide decisions as to whether a grief reaction can meet criteria for major depression. It stated: ‘The normal and expected response to an event involving significant loss (e.g. bereavement, financial ruin, natural disaster), including feelings of intense sadness, ruminations about the loss, insomnia and weight loss, may resemble a depressive episode. The presence of symptoms such as feelings of worthlessness, suicidal ideas (as distinct from wanting to join a deceased loved one), psychomotor retardation, and severe impairment of overall function suggest the presence of a Major Depressive Episode in addition to the normal response to a significant loss’. The first sentence expands the definition of ‘loss’ beyond bereavement, and suggests that any such losses may induce normative responses or ones that resemble a ‘depressive episode.’ The second sentence is close to DSM-IV's Criterion E but is tentative (e.g. ‘symptoms such as’, ‘suggest’) and both contribute to a rather ambiguous tone.

The slide to similarities

By failing to provide criteria differentiating bereavement and major depression, and by more emphasizing the similarities (i.e. shared life event stressors, depressive symptoms), the DSM-5′s website note's [7] sub-text appears to still position bereavement as paralleling – or commonly associated with – major depression. In essence, factors common to bereavement and major depression are detailed rather than factors that differentiate them. This essay will consider evidence of the latter, and some implications of the current DSM-5 model and its debate.

Overstraining stress

As quite contrasting psychiatric conditions can be preceded by similar stressors, clinical distinction and diagnostic status is rarely argued by a stressor specificity model or rejected by stressor commonality. Differentiation of psychological normative states from psychiatric conditions – as well as differentiation of the latter conditions between themselves – is generally better advanced by considering phenomenology, natural history and treatment specificity.

Differences in phenomenology

What has been generally missing from the multiple commentaries on the DSM-5 classification – with a distinct exception being a recent paper by Shear [8] – is any detailed phenomenological distinction between grief and depression, rather than reference to their commonality symptoms (such as emptiness, crying, appetite and sleep disturbance). In reporting their classic study of ‘normal bereavement’, Clayton et al. [9] referred to Freud's 1917 [1] differentiation of grief (normative, self-limiting and with ‘absence of disturbances of self-regard’) from depression (which was weighted to ‘disturbances of self-regard, self-condemnation’). The latter construct of decreased self-esteem and of increased self-criticism is central to definitions of depression provided by Grinker [10], Beck [11], Becker [12], and Engel [13], and intrinsic to many depression measures, including the General Health Questionnaire [14] and the Depression Anxiety and Stress Scale [15]. Clayton and colleagues [9] emphasized that ‘disturbances of self-regard, self-condemnation and suicidal thoughts’ were absent in their naturalistic study of bereaved relatives. By contrast, a depressed state was confirmed and defined quite parsimoniously in early factor analytic studies [10] by central constructs of lowered self-esteem and increased self-criticism. Shear [8] enriches those points of distinction in observing that the hallmark of grief is a blend of yearning and sadness, along with thoughts, memories and images of the deceased person, while in contrast, depressed people ‘see themselves and/or the world as fundamentally flawed, inadequate or worthless.’ In essence, the psychological pain in ‘normative grief’ emerges from loss of the ‘other’ – and self-esteem is almost invariably preserved in the early stages – while the central characteristic of depressed states is compromised self-worth. The phenomenological distinction is sharp.

Differences in natural history

Natural history has also been only minimally considered in the recent commentaries. Episode recurrence may provide a point of distinction. Wakefield and Schmitz [16] detailed two studies establishing that the recurrence rate of major depression in those with bereavement-related depression was no different from those in the general population with no history of major depression, and noted that those with bereavement-related episodes were less likely to have impaired role functioning, psychiatric treatment, or comorbid disorders. In a study by Mojtabai [17], not only were those with a bereavement-related, single, brief depressive history at no greater risk of a depressive episode than those in the general population, but their bereavement episode (compared to those with bereavement-unrelated brief depressive episodes) was associated with less impairment, a lower prevalence of comorbid anxiety and less treatment seeking. The author concluded that eliminating the DSM-IV bereavement exclusion for major depression risked expanding the definition of major depression to include ‘emotional reactions to loss that are self-limiting, and not associated with future risk of depression’.

Differences in staging

Staging is another important distinction. While clinical depression may be presaged by warning signs or symptoms, and it may have a slow or abrupt onset, it generally lacks the stages integral to grief. As detailed by Parkes [18], grief has a first stage manifested by the shock of realization and a sense of numbness; a second-stage alarm state with marked physiological arousal compromising sleep and appetite; a third-stage search urge (with the individuals searching and pining for the lost object); and a fourth stage of anger, guilt or internal loss (perhaps reflecting the bereavement but also the impacts of loneliness, loss of a role, financial and other consequences); before a fifth level, resolution, may be achieved. The initial phases more reflect ‘separation anxiety and distress’ in relation to the loss of another and without diminution of self-esteem. In a more recent empirical study of grief patterns, Maciejewski and colleagues [19] found ‘at least partial support for the stage theory of grief’, identifying five stages of disbelief, yearning, anger, depression and acceptance, but observed that ‘yearning, not depressive mood, is the salient psychological response to natural death’. The atavistic nature of such attachment loss responses is well illustrated in animal studies which hint at evolutionary advantages – at least in the young.

For example, Kaufman and Rosenblum [20] reported studies of infant pigtail monkeys separated from their mother, and detailed three principal stages. The first phase was one of highly ‘agitated’ behavior, with the infant macaque running around in a seeming erratic manner and emitting high-pitched screams. In the second phase it assumed a slumped position and made no sound. The third phase was described as one of ‘recovery’ in that there was a gradual increase in movement and interaction with the social environment. The evolutionary advantage to the ‘agitation’ phase is that, if the macaque's mother is in proximity, she and her infant are more likely to be reunited. However, if she is missing or taken by a predator, the infant's second phase of ‘conservation withdrawal’ behaviours is advantageous in protecting it from being observed or heard by any predator, while its slumped posture protects it against dehydration and heat loss.

Humans sometimes evidence such distinctively overt phases, perhaps most gravidly illustrated by women who experience the death of their baby. In essence, grief evidences a distinctive set of phases in response to a break in an attachment bond. As detailed by Bowlby [21] and summarized by Rutter [22] ‘Bond disruption is the key variable’ to a grief process. Its commonality to many so-called ‘reactive depressive’ states will be considered shortly, together with implications.

Differences in treatment response

Turning to treatment specificity, Shear [8] expressed a common argument in stating that ‘depression requires treatment and grief requires reassurance and support’. The evidence base for antidepressant medication is convincing in relation to major depression – but is limited for the management of grief and often contingent on other factors, predictably including the presence or absence of a superimposed depression. For example, Bryant [2] observed that, while grief reactions do not respond to antidepressant drugs, ‘bereavement-related depression responds to antidepressant interventions.’ The predictable difficulty in seeking to clarify evidence of the latter is whether any such responsivity emerges from the depression per se or from some components of the bereavement-depression complex. Further, as depressed and depression-prone individuals are not immune to bereavement and the experience of grief, does any such responsivity emerge from such independent factors or from interdependent ones?

Kendler's contentions

But let us return to the DSM-5 Mood Work Group's key argument for removing the bereavement criterion from the definition of major depression. Kendler [5] stated – building on an earlier publication [23] - that there is little difference between major depression in response to bereavement compared to response to other stressors (e.g. being physically assaulted or raped, being betrayed by a trusted spouse, learning that a loved child is dealing drugs, learning that one has cancer or losing a treasured job). Further, he observed, ‘…the DSM-IV position is not logically defensible. Either the grief exclusion criterion needs to be eliminated or extended so that no depression that arises in the setting of adversity would be diagnosable.’

Let us pursue his second exemplar via a commonly observed clinical vignette – a woman distressed after being told by her partner/spouse that he is leaving her for another. Over the first week or two she sleeps minimally, has no appetite, loses weight rapidly and is hypervigilant (e.g. thinking she sees him in a crowd or his car as she is driving home). She is desperate to be reunited with him, and has to be encouraged by friends not to search for him or to make any contact. After several weeks, her physiological arousal settles, and she appears more withdrawn and preoccupied. Such a biphasic ‘bond disruption’ response is akin to the macaque exemplar provided earlier – separation anxiety followed by conservation withdrawal. During or following that second phase, if she comes to the view that her ex-partner was a total ‘jerk’, there may be no impact on her self-esteem. By contrast, if she judges that the relationship break-up proves her inability to maintain relationships or that she is of so little worth that no one will ever love her or seek her company, she is likely to feel and experience a depressed mood and even a clinical depressive episode. Her interpretation of the event – rather than the event itself – may alone shape the likelihood of the response and the ‘syndrome’: ‘grief’, or a ‘reactive depression’ with compromised self-esteem.

Do the types of loss broaden the territory of grief?

Wakefield [24] observed that depressive symptoms are common in response to losses other than bereavement and that they ‘divide into more benign, seemingly normal-range misery’ and more substantive disordered responses’. Why? Let us consider the other stressors nominated by Kendler [5]. Being assaulted is an affront to one's safety and privacy, learning that one's child is dealing drugs may be processed as a loss of an ideal, learning that one has cancer may be loss of one's hopes for immortality and losing a ‘treasured job’ might well invoke a loss of one's professional identity. Some involve a ‘bond disruption’ but certainly all involve a loss. Dependent on whether the individual's attributional processing of the loss compromises their self-esteem – each may also induce a reactive depressive state or condition.

As noted, Kendler provides only two polarized options – but in life and in the ineffable world of Psychiatry there are always multiple options. Kendler's second option (i.e. no depression arising following adversity would be diagnosable) is at variance with Psychiatry's overall paradigm of its conditions being diathesis-stress disorders, and with some conditions weighting a greater contribution of the former and others weighting the latter. To argue that any depression preceded by any adversity would not be diagnosable would not respect clinical reality. But in comfortably rejecting that option it does not mean that we are required to assume Kendler's alternative option in his binary model (i.e. the DSM-IV grief exclusion criterion requires removal). A third option raised by such considerations is to consider the extent to which Kendler's stress exemplars might actually broaden the territory of ‘grief.’

And so Pandora's box is further opened. Rather than drawing bereavement within the domain of the clinical depressive disorders (as DSM-5 appears still to favour), we might better lean the other way and consider whether many currently positioned clinical depressive disorders (especially the reactive depressive conditions) might fit more comfortably within a grief paradigm, and benefit more from management weighting such a model.


Many have put a reductionist view that grief is completely normative and depression is always clinical, and that any merging will ensure a wide-scale move to manage bereaved people with psychotropic medication. To be fair, DSM-5 is likely to be addressing a reality – that some bereaved individuals will develop a clinical depressive episode – and that their treatment might then involve a differing management paradigm to that for normative grief. However, the recent DSM-5 annotation or website note [7] appears underpinned by Kendler's observation [5] that the DSM-IV grief exclusion criterion is ‘not logically defensible’.

This essay seeks to broaden the list of logical parameters that invite consideration. In particular, it invites weighting of phenomenological distinction between grief and depression, and consideration of an alternate paradigm that allows some reactive depressive disorders to be more logically modeled and managed as ‘grief’ states.


This report was funded by an NHMRC Program Grant (510135).

Declaration of interest