What should ICD-11 do that DSM-5 did not?

ICD Insights

There are known knowns; these are things we know that we know. There are known unknowns; that is to say, there are things that we now know we don’t know. But there are also unknown unknowns – there are things we do not know we don’t know.

—Donald Rumsfeld, former US Secretary of Defense

Like monozygotic twins reared apart, the Diagnostic and Statistical Manual of Mental Disorders (DSM) and International Classification of Diseases (ICD) exist independently yet share common heritage in their make-up and structure. Both aim to identify and operationalize the entirety of psychiatric disorders. The ICD has long been the official worldwide diagnostic coding system, while the DSM elaborates more descriptive detail and exclusionary criteria. Each new iteration of either document has been accompanied by calls for greater synchrony between them. Notably, while most differences between ICD-10 and DSM-IV involved nuanced variations in phrasing, First (2009) observed that 21% of the disorders found in both systems involved substantive conceptual differences. (For example, ICD-10, unlike DSM-IV, doesn’t differentiate bipolar I from bipolar II disorder; or, schizoaffective disorder in ICD-10 requires a balance in the severity and duration of psychosis versus mood symptoms, while DSM-IV requires 2 weeks of psychosis alone, plus the existence of mood symptoms for a ‘substantial portion of the total duration’ of illness.) To reflect upon the question of what ICD-11 should do that DSM-5 didn’t – particularly in the wake of much-publicized controversy about the whole DSM-5 process (Frances, 2013) – it might first be useful to ask what, if anything, DSM-5 did (or shouldn’t have done) that DSM-IV didn’t.

At the outset, one must acknowledge that both the ICD and DSM focus more on the reliability than validity of the disorders they describe. Hence, several proposed DSM-5 constructs that seemed either scientifically intriguing (e.g. psychosis risk syndrome) or reflective of what real-world practitioners see (e.g. mixed anxiety-depressive disorder) were omitted because they yielded poor inter-rater reliabilities in field trials (Regier et al., 2013). Superficially, one could nitpick about semantic changes in DSM-5 (e.g. ‘adjustment disorders’ are now ‘stress response syndromes’) or finer points that reflect arbitrary consensus opinions (such as whether attention-deficit/hyperactivity disorder more accurately begins before age 12 than before age 7), sometimes despite empirical evidence that contradicts consensus (such as observations that fewer than 4 days’ duration likely constitutes a hypomanic episode (Angst et al., 2012)). More philosophical rumblings come from debate about whether or not DSM-5 was right to eliminate bereavement as an exclusion criterion for diagnosing major depression – a move divergent from the perspective of the Chair of the World Health Organization Working Group on the Classification of Mood and Anxiety Disorders (Maj, 2012); or to dismantle DSM’s multiaxial system; or to have made obsessive-compulsive disorder a free-standing construct separate from anxiety disorders; or to devise new entities such as ‘disruptive mood dysregulation disorder’ to describe youth with affective instability and externalizing behaviors.

DSM-5 has lumped more than split diagnostic hairs, as reflected in its reclassification of conditions such as hypochondriasis or somatoform pain disorders under the broader rubric of ‘somatic symptom disorders’; or, the subsuming of substance abuse and dependence simply as ‘substance use disorders’ – but with no scientific acknowledgement of addiction as a core process, irrespective of pharmacologic corollaries such as tolerance and withdrawal. Developmentalists will either hail or lament DSM-5’s elimination of Asperger’s syndrome alongside broadening of the autism spectrum disorders construct. Should ICD-11 follow suit purely for the sake of harmonization, if not scientific accord?

Some critics have argued that across editions of the DSM, criteria for certain syndromes have become overspecified, to the point where differential diagnostic features become impractical if not meaningless for the average practitioner. For example, when using the new DSM-5 mood disorders ‘mixed features’ specifier, will clinicians really labor to determine if hypomania criteria are met during a depressive syndrome only after discounting the four polarity-overlapping features of irritability, insomnia, distractibility and indecisiveness (Malhi, 2013)? And will applying this specifier to unipolar depressed patients merely encourage a ‘wastebasket’ approach to diagnosing mixed states in anyone, without bothering to determine if a true manic or hypomanic episode ever occurred?

One wonders also about the prevalence and meaningfulness of diagnoses labeled as ‘not otherwise specified (NOS)’ (or, as renamed in DSM-5, ‘not elsewhere classified’). Research diagnosticians invoke DSM’s NOS/NEC terminology when a patient’s presentation falls shy of threshold criteria by one or more symptoms, or is the minimum days’ duration. Non-academic clinicians tend to care less about criterion-counting, and (one might dare say) make NOS diagnoses as non-committal or even hasty place-holder labels for patients with complex features. Should ICD-11 elaborate further on how best to classify forme fruste or other ‘atypical’ presentations that genuinely fail to conform neatly to operational criteria?

It’s easy enough to express dissatisfaction with how DSM-5 mainly reshuffled the phenomenologic deck without incorporating much, if any, new science to inform diagnosis, a fate ICD-11 will likely avoid no more successfully. Indeed, many critics have disparaged the timing of a new DSM as being far too premature to integrate findings from genetics or clinical neuroscience as a means of diagnostic refinement. Perhaps the most extreme disparagement came from the decision by National Institute of Mental Health (NIMH) leadership to eliminate DSM-5 clinical diagnoses altogether in favor of the Research Domain Criteria (RDoC; www.nimh.nih.gov/research-priorities/rdoc/nimh-research-domain-criteria-rdoc.shtml), reducing all of mental illness to a handful of broad putative phenotypes thought to map more directly to neural circuits. While RDoC does highlight the concept of dimensions rather than categories of brain function and dysfunction – a goal originally espoused as an intended innovation of DSM-5 (Kraemer, 2007) – it also creates a formidable divide between clinicians and investigators. Herein lies the major question of whether the language of nosology is spoken differently by researchers than practitioners. No iteration of either DSM or ICD has acknowledged this fundamental distinction about who uses diagnostic classifications, and for what purpose?

Philosophically devised diagnoses come and go (remember homosexuality in DSM-II?). Until prospective follow-up studies document the longitudinal stability of newly proposed diagnoses or criteria sets (or, until true biomarkers exist), their construct validity remains provisional. Will ICD-11 yield to new diagnostic headings based only on their popularity rather than construct validity (such as, ‘soft spectrum bipolar disorders’)? Or, more wisely, will it change little in the absence of etiologic or other diagnostically meaningful data? Perhaps neither the ICD nor DSM should do anything astray from one another, or from their own earlier incarnations, until at least the next millennium, or the publication of even one robust genome-wide association study for any disorder – whichever comes first!