Differential diagnosis of depressive illness versus intense normal sadness: how significant is the ‘clinical significance criterion’ for major depression?

Failure of the major depression clinical significance criterion to eliminate false-positives

As psychiatry has moved from the asylum to the community over the last century, and embraced operationalized descriptive diagnostic criteria, the distinction between nonpsychotic unipolar depressive disorder (‘major depression’) and intense normal sadness in response to loss has proven difficult to formulate in terms of symptomatic criteria. However, this distinction is important for prognosis and treatment choice.

According to the American Psychiatric Association’s Diagnostic and Statistical Manual of Mental Disorders Fourth Edition (DSM-IV) [1], the diagnosis of major depression requires the presence for 2 weeks of at least five symptoms out of a list of nine that includes, for example, sadness, loss of interest in usual activities, lowered appetite, fatigue and insomnia. However, these symptoms also occur in intense normal emotional responses to loss. False-positive diagnoses occur when such normal sadness responses are misdiagnosed as major depressive disorder [2]. Recent studies suggest that most, and perhaps almost all, individuals have five such symptoms for 2 weeks or longer at some point in their lives in response to loss [3]. Therefore, the question arises: how many of these individuals are afflicted by mental disorder and how many are responding within normal limits and simply emotionally processing their changed circumstances?

In an effort to reduce false-positives [4], a clinical significance criterion (CSC) requiring distress or role impairment (i.e., “the disturbance causes clinically significant distress or impairment in social, occupational or other important areas of functioning”) was added to those DSM-IV diagnostic criteria sets, including major depression, in which “the symptomatic presentation by itself (particularly in its milder forms) is not inherently pathological and may be encountered in individuals for whom a diagnosis of mental disorder is inappropriate” [1]. The CSC has been the primary way by which the DSM attempts to prevent false-positives. Subsyndromal depression, which can be diagnosed with as few as two symptoms, also requires satisfaction of the CSC. However, the effectiveness of the CSC remains unknown.

In a recent study, we asked: “does the DSM-IV clinical significance criterion for major depression reduce false-positives?” [5]. Using evidence from the National Comorbidity Survey Replication (NCS-R) [6], a nationally representative community survey of mental disorder using DSM-IV diagnostic criteria, we examined whether the CSC performed its intended function of reducing community rates of diagnosis by eliminating substantial numbers of symptomatic cases that do not include distress or impairment from the disorder category.

Our empirical study tested a prediction that one of us (Wakefield), with coauthor Robert Spitzer (the originator of modern operational diagnostic criteria), had made in a conceptual article a decade before [2]. We argued for the ‘redundancy thesis’ that the CSC would not eliminate false-positive diagnoses of major depression, because anyone having the specified symptoms – even if experiencing a normal intense reaction to loss – would likely experience distress (which is inherent to sadness) or role impairment (because sad people tend to withdraw and show less energy for usual tasks). Thus, the CSC was redundant with the symptom criteria and could not distinguish normal from disordered symptoms.

That is exactly what our study found [5]. The CSC eliminated very few cases. Even subsyndromal cases with minimal symptoms included distress or impairment in almost all instances. Of 2071 individuals who reported persistent sadness, 97% satisfied CSC criteria for distress. Moreover, of 817 individuals who did not qualify for major depression, mostly because they had too few symptoms, 94% reported significant distress. Of 1542 individuals who satisfied depression symptom and duration criteria, 96% satisfied criteria for impairment in role functioning. Indeed, despite the fact that the CSC was added to the NCS-R and had not been included in the original NCS [7], NCS-R lifetime adult depression prevalence rates increased.

What this means at the clinical level is that after ascertaining the presence of depressive symptoms from a patient, going on to ask about clinical significance in terms of distress and role impairment – while important for the clinician’s understanding of the nature of the patient’s condition – is unlikely to help differentiate normal versus disordered sadness. For screening and epidemiologic instruments, this means a savings in time and money without a loss in validity.

However, this leaves the false-positive problem unaddressed. Moreover, the CSC, while useful in excluding extremely mild cases from diagnosis, is flawed in principle as a general solution to the false-positives problem, because it essentially measures the harm from a condition, but normal distress can be harmful too. To judge a condition a disorder, according to the DSM’s definition of mental disorder, one must also infer that it is due to an internal dysfunction – that is, something must be going wrong with biologically designed processes within the individual [8,9].

The essential problem is that the DSM criteria do not place the symptoms into their context. Traditionally, given the lack of understanding of brain mechanisms, the primary way psychiatric medicine has been able to discriminate normal reactions from symptoms due to internal dysfunctions is via a reference to context. The reason is simple: human emotions are largely designed to be triggered by certain contexts and not by others, so without reference to the context, one cannot tell whether emotions are normal or disordered [10].

Independently of our study’s findings, the DSM-5 Task Force is considering removing the CSC from diagnostic criteria and placing it in an additional diagnostic axis [11]. One reason for this is to bring the DSM into conformity with WHO’s International Classification of Diseases [12]. Another reason is that, when there are potential comorbid psychiatric disorders, the CSC is difficult to apply because the clinician must judge which (one or both) of the sets of symptoms are responsible for the patient’s distress or role impairment.

This reminds us of the special problems for the CSC’s application within the context of neurological disorders. The distress and role impairment – often profound – that accompany some neurological conditions can make application of the CSC all but impossible. In addition, some of the more extreme symptoms that might be used to distinguish normal sadness from clinical depression – such as the presence of psychomotor retardation – may be difficult to evaluate with some neurological patients.

Normal sadness versus depressive disorder in neurological disease

The distinction between normal sadness and depressive disorder, often under other labels, such as ‘reactive depression’, has long been part of the neurologist’s concerns. Applying clinical common sense, neurologists have generally looked to the context of depressive symptoms to try to make this distinction.

In his classic paper, ‘Depression as viewed through neurologic spectacles’, Nicol asserted that the neurologist “must be alert in order to recognize depressive illness when it occurs, and at the same time be thoroughly familiar with conditions that mimic it” [13]. And, one might add, today we are also aware of the crucial problem of depression, as well as sadness and sheer weariness, in the patient’s caregiver [14]. In both cases, individuals may be reacting normally to terrible circumstances.

The conditions that mimic depressive illness that Nicol primarily had in mind consisted of neurologic diseases that have depressive symptoms as a direct physiological effect, neurologic diseases that in their clinical course yield emotional lability or apathy that can be mistaken for physiologically induced depression, and the depressive side effects of some medications commonly used to treat neurological disease. Such depressive conditions are segregated from major depressive illness by the DSM and diagnosed under mood disorder due to a general medical condition or to a substance.

But what of the psychological ‘mimics’ of depressive disorder? From Hippocrates to Kraepelin, the standard view in medicine was that the single greatest mimic of depressive disorder is intense normal sadness brought about by loss [10]. The physician Aretaeus of Cappadocia (ca. AD 150–200) recorded one of the first false-positives of this kind, in which the symptoms of unrequited love were mistaken for melancholia [15].

The distinction is important in the neurological context where what passes as depressive disorder may be despair. Using the term ‘reactive depression’ for what is often an intense normal response, Nicol wrote: “It should surprise no one that depression, particularly the reactive form, accompanies organic central nervous system and neuromuscular disease. The unfavorable prognosis so often associated with these maladies often makes it necessary for the attending physician to sufficiently disguise his diagnosis so as to avoid precipitating a reactive depression” [13]. Moreover, neuromuscular diseases that “render a previously muscular individual powerless” [13] may cause reactive depression that can be addressed through specific therapies, Nicol argues. Whereas such conditions may trigger a true pathological depression, the initial depressive reaction in such cases, one might argue, is often proportionate to the magnitude of the loss and the need to process it – the classic sign of sadness that is a normal part of human nature and more likely to have a benign trajectory and be treatable by psychosocial means.

Studies of treatment of depression occurring in the immediate aftermath of a stroke show very high spontaneous remission/placebo improvement rates comparable with medication, suggesting that some instances of such depressions may be intense normal reactions to a dire situation with a trajectory of remission and resilience characteristic of normal sadness [16,17]. Other studies suggest that depressive symptoms in multiple sclerosis patients may be more related to their perception of the disease’s prognosis than to the actual disability they are experiencing or the actual prognosis [18,19]. Consequently, adequate differentiation of normal sadness versus depressive disorder could suggest interventions, such as education or psychotherapy. Some antidepressants are contraindicated in some neurological disorders, and in general the burden of multiple medications and their potential side effects and interactions suggests that the appropriateness of alternative treatments is a factor to which the clinician should be alert. Such alternatives are particularly available in nondisordered reactions of mourning real losses.

Sobel et al. note that studies suggest that depression in multiple sclerosis may be related to the patient’s reaction to environmental and social factors [20]. In particular, it appears to be the rapidity of changes in level of disability and the patient’s inability to accept and adapt to the changing circumstances, more than level of functioning, that is associated with depressive symptoms. They emphasize that major depression “often is a reaction to the toll of disability on the patients and caregivers” [20]. Although lesion location, volume and other brain-structural features explain a substantial amount of the variance in depression in multiple sclerosis patients, psychosocial variables explain about the same amount of the variance [21].

It is not uncommon to interpret end-stage patients’ wishes to die as the expression of depression, and straightforward application of DSM criteria might confirm this. But further exploration suggests that sometimes it is the confounding of depressive symptomatology with the wish to die via DSM’s ‘death-related and suicidal thoughts’ criterion that is partially responsible. One can easily confuse a rational, deliberatively formulated wish to die in end-stage neurological disease with a depressive symptom [22]. Moreover, as one might expect, neurological patients suffering chronic pain are more likely to be depressed [23]. Automatically labeling these various circumstances of the patient – to which human responses of sadness are natural – as ‘risk factors for depressive disorder’ translates human meaning into a medical idiom that can be more misleading than enlightening.

Conclusion

For neurologists, the DSM ‘symptom checklist’ approach to depression diagnosis is particularly problematic because neurological patients’ depressive symptoms are often commingled with similar neurological symptoms. The neurologist faces a particular challenge in assessing depressive symptomatology: is it normal despair over the medical condition, part of the course of the neurological condition anchored in physiology, a side effect of the medication or a genuine additional major depressive disorder? Our research suggests that whether in neurology, general medical practice or psychiatry, the CSC does not help in understanding whether the symptoms represent mourning or melancholia. We believe it is time for the DSM to consider new ways to help clinicians make this diagnostic discrimination, including the greater incorporation of contextual considerations into diagnostic criteria.

Nicol explained at the end of his paper that “I have tried…to caution you against making a too hasty diagnosis of depression in the presence of signs that mimic it” [13]. Both the mental health professional and the neurologist must keep in mind that intense normal sadness is the most common condition that mimics depressive illness. Approaching the patient as a person who retains the range of normal human emotion, including sadness, hopelessness and despair, in response to contextual circumstances that may be dire, potentially opens novel pathways to treatment or to constructively engaging the patient and the caregiver.

Financial & competing interests disclosure

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.