Challenging use of the insight specifier for body dysmorphic disorder

In the latest edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5), body dysmorphic disorder (BDD) is characterised by preoccupation(s) with perceived flaws in physical appearance that are not visible or appear slight to others, alongside repetitive behaviours or mental acts that serve to (temporarily) ameliorate said appearance-related concerns (American Psychiatric Association, 2013). A similar definition was adopted in the updated International Classification of Diseases (ICD), 11th Revision, with added mention of experiences of excessive self-consciousness alongside ideas of reference (World Health Organization, 2018). Across both classification systems, the nosological status of BDD has been challenging and fraught with inconsistencies. It was first conceived as ‘dysmorphophobia’ by Italian physician, Enrico Morselli, in the late 19th century and then described as a compulsive neurosis and obsessional shame about the body by early psychiatrists, Emil Kraepelin and Pierre Janet, but only officially introduced into the 1987 and 1979 editions of Diagnostic and Statistical Manual of Mental Disorders, Third Edition, Revised (DSM-III-R) and ICD-9, respectively. At that point, its initial monikers associated with hypochondriasis (e.g. ‘beauty hypochondria’) were likely instrumental in linking it among the somatoform disorders.

A prominent feature of BDD is that the appearance-related preoccupations often reach delusion proportions. In fact, a high proportion of patients with BDD demonstrate delusional conviction and referential delusions, but insight can vary. In ensuing iterations of its nosological classification, the assignment of non-delusional BDD within the somatoform disorders remained, but uniquely its delusional variant was separately classified as a psychotic disorder (i.e. delusional disorder, somatic subtype), with a double-coding approach subsequently required in Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision (DSM-IV-TR). The approach in ICD-10 was even more complex, with BDD falling into multiple possible diagnoses involving hypochondriacal disorder, schizotypal disorder or (other persistent) delusional disorder (Veale and Matsunaga, 2014). Recent updates to these classification systems now designate BDD as a standalone diagnosis, but grouped within the obsessive-compulsive and related disorders (OCRDs) in DSM-5 as well as in ICD-11. Along with this move came the recognition that its non-delusional and delusional variants constitute a single disorder differentiated along a spectrum of insight. Consequently, an insight specifier was introduced – this comprises three levels (good or fair, poor, or absent/delusional beliefs) in DSM-5 and two levels (fair to good or poor to absent) in ICD-11. Yet there is reason to question this approach on three fronts: (1) from a theoretical perspective, (2) what this means for nosology and (3) its impact on clinical utility in terms of treatment.

First, the prevailing model of multidimensional insight theorises this construct as comprising three discrete facets involving the recognition that one has a mental illness, recognises the attribution of unusual mental events as pathological and accepts the need for treatment (David, 1990). In contrast, the insight specifier for BDD is seemingly geared towards assessing level of conviction; individuals with fair insight tend to embrace overvalued ideas susceptible to disconfirmation, whereas those endorsing poor insight hold extreme, unwavering conviction. From this perspective, more accurate terminology would perhaps evoke use of a conviction specifier. The nature of delusional beliefs also warrants further consideration. A multidimensional assessment of insight based on principal preoccupation showed that levels of conviction and fixity of ideas in BDD were significantly higher than those in obsessive-compulsive disorder (OCD) and were in fact similar to schizophrenia (Toh et al., 2017). On the contrary, what was termed insight, or the ability to attribute beliefs to a psychological cause (David, 1990), was significantly lower in BDD relative to people with OCD or schizophrenia. Furthermore, referential delusions (alluded to in ICD-11 but not specifically mentioned in DSM-5) were significantly elevated in BDD compared with OCD, and even schizophrenia. Other dimensions worth considering in an assessment of insight include the role of emotion or contribution of overvalued ideas (as opposed to delusions per se).

Second, use of the insight specifier cuts across the OCRDs in ICD-11, but is constrained to only three disorders – BDD, OCD and hoarding disorder in DSM-5. This engenders two key problems: (1) management of insight in disorders where its absence is a key feature, and (2) omission of an insight specifier for other disorders where delusional beliefs are known to be prominent. For instance, in the schizophrenia spectrum disorders, lack of insight is ‘typically a symptom of schizophrenia itself’ (American Psychiatric Association, 2013: 101) and underlies key diagnostic criteria involving delusions and hallucinations. Yet in schizophrenia, affected persons tend to exhibit varying levels of insight that can fluctuate, depending on illness course or severity and other concomitant factors. We question why an insight specifier is not routinely employed within this class of disorders where impaired insight significantly predicts treatment non-adherence, heightened relapse rates and other adverse outcomes (American Psychiatric Association, 2013). There is also an apparent lack of consistency in deployment of an insight specifier across other classes of disorders. For instance, up to half of the individuals with anorexia nervosa endorse delusional eating-related beliefs, although there is no insight specifier in its diagnostic criteria. Therefore, despite greater uniformity in applying the insight specifier within all OCRDs in ICD-11, there is still a persistent lack of consistency in its use across disorder classes for both classification systems. Addressing this limitation in future editions of DSM and ICD will aid in advancing psychiatric nosology and clinical and research efforts.

Third, the clinical utility of having an insight specifier for BDD is questionable. This is partly because insight in BDD is not routinely assessed in clinical settings, and more importantly, there seems to be a lack of guidelines in terms of how this should best be done. At present, both DSM-5 and ICD-11 diagnostic criteria seem to advocate an appraisal of conviction levels, but the multidimensional nature of insight implies that other facets of this construct (e.g. fixity of ideas, treatment adherence) might be more relevant for clinical decision-making, in terms of optimum choice of intervention (e.g. cognitive-behavioural therapy vs psychopharmacology). Related to this, the insight specifier does not guide use of antipsychotics versus selective serotonin reuptake inhibitors (SSRIs), both of which can be somewhat effective for delusional (or non-delusional) BDD patients. There is also a risk that clinicians may employ the insight specifier as a proxy measure of BDD severity (and/or treatment engagement), which could be misleading. Until a more accurate assessment of multidimensional insight may be incorporated within these classification systems, there thus seems to be little clinical utility in advocating for use in its current manifestation.

In conclusion, a theoretical case could be made for the use of an insight specifier in BDD on the one hand, namely, a considerable percentage of affected persons do endorse appearance-related beliefs that are of delusional tenacity, and this has significant impact on illness outcomes. However, closer examination of the specific nature of delusional beliefs in the disorder shows that these may not be solely restricted to appearance-related concerns. Where they occur, the severity and impact of these delusions are notably at least equal to those observed in schizophrenia (Toh et al., 2017). Conversely, there seems to be limited clinical utility advocating for an insight specifier in its present form. More importantly, before it can be constructively adopted in clinical settings, additional clarity needs to be provided in terms of its specific operationalisation, alongside concrete guidelines delineating how it should be assessed and what the treatment implications would likely be.