Historical framework for psychiatric nosology

In the review, Kendler (pp. 1935–1941) outlines an historical framework for ongoing efforts to develop a scientific psychiatric nosology. Drawing parallels with the history of biological taxonomy, which began with ‘expert’ classifications and has moved on to the use of a plurality of ‘bottom-up’ approaches, Kendler notes how psychiatry is moving from the ‘expert’ opinions on which classifications were initially based to approaches based on a variety of illness characteristics. In these efforts, given the historically contingent nature of classificatory systems, Kendler argues that advances can be best achieved through a process of epistemic iteration, in which each revision can be expected to improve the performance of the nosology.

Regular papers

Four papers examine aspects of psychosis, post-traumatic stress disorder and borderline personality disorder. In the first, Demjaha et al. (pp. 1943–1955) investigated symptom dimensions, and their associations with risk indicators and clinical variables including diagnosis, in a sample of 536 individuals with a first episode of psychosis. In a factor analysis of data collected using the Schedules for Clinical Assessment in Neuropsychiatry, the authors found that a five-factor solution best fit the data, revealing manic, depressive, reality distortion, negative and disorganization symptom dimensions. The authors further found that specific dimensions were correlated with specific variables (e.g. manic symptoms with short duration of psychosis, acute mode of onset, etc.). In addition, the authors found that combining dimensions with diagnosis explained the most variance in sociodemographic and clinical characteristics.

Elhai et al. (pp. 1957–1966) examined the impact of using alternate structural models for the diagnosis of lifetime post-traumatic stress disorder (PTSD) while retaining the six-symptom diagnostic requirement for PTSD in DSM-IV using data from the US National Comorbidity Survey Replication and the US National Survey of Adolescents. In comparisons of empirically supported four-factor models of PTSD with the DSM three-factor model, the authors found that the diagnostic alterations resulted in substantially improved structural validity, a decrease of 1–2.5% in estimates of prevalence, and equivalent associations with comorbidity and sociodemographic variables.

Aggen et al. (pp. 1967–1978) investigated the measurement invariance of borderline personality disorder (BPD) (i.e. whether criteria assess BPD similarly across groups) in a sample of 2794 Norwegian twins. In analyses based on item-response modelling, the authors found evidence that the DSM-IV BPD ‘impulsivity’ and ‘affective instability’ criteria function differently with regard to age and sex, this being most marked for ‘impulsivity’. The authors comment that, if replicated, these findings may have implications for the interpretation of prior research that has used these criteria.

Fertuck et al. (pp. 1979–1988) examined capacity to discriminate the mental state of others from expressions in the eye region of the face in a sample of 30 individuals with BPD and 25 controls, using the Reading the Mind in the Eyes Test (RMET). The authors found that those with BPD performed better than controls on the RMET independent of potential confounders, particularly for the Total Score and Neutral RMET performance. The authors conclude that enhanced sensitivity to the mental states of others may be the basis for social impairments in BPD.

A meta-structure for DSM-V and ICD-11

Following a brief Introduction by Kendler, Andrews et al. (pp. 1993–2000) begin this series of papers by outlining a proposed meta-structure for DSM-V and ICD-11 based on both risk and clinical factors. Andrews et al. began this process by allocating most DSM-IV disorders to one of five clusters. Following literature reviews by teams of experts to determine within-cluster similarities on 11 predetermined validating criteria, the final proposed clusters that shared risk and clinical factors were: neurocognitive, neurodevelopmental, psychosis, emotional and externalizing. The following five papers present findings of the reviews for each of these proposed clusters.

Sachdev et al. (pp. 2001–2012) assessed the validity of the neurocognitive cluster. The authors note that what distinguishes this cluster from other disorders is the salience of cognitive deficits and evidence for demonstrable neural substrate abnormalities. The occurrence of disorder subsequent to normal brain development sets this cluster apart from the neurodevelopmental cluster. With regard to aetiology, the neurobiological underpinnings are better understood than for other disorders. However, the authors note that there is less consistent evidence concerning shared biomarkers, co-morbidity and course.

Andrews et al. (pp. 2013–2023) assessed the validity of the neurodevelopmental cluster. The authors note that this cluster comprises a heterogeneous group of disorders (e.g. Mental Retardation; Learning, Motor and Communication Disorders) and is distinct from ‘childhood’ disorders. There are some shared risk and clinical factors, including evidence of a neurodevelopmental genetic phenotype, salient cognitive symptoms and an early and emerging course. The authors further note that other childhood disorders share similarities with externalizing and emotional clusters, and in the proposed meta-structure these are allocated to a ‘not yet assigned’ group.

Carpenter et al. (pp. 2025–2042) assessed the validity of the psychosis cluster. The authors note that this group is consistent at the level of shared psychotic psychopathology, with the exception of bipolar disorder (BD) and schizotypal personality disorders (SPD). The authors further note that there is modest overlap between schizophrenia and BD in risk factors, cognition and endophenotypes, but also key differences. There is more evidence for a spectrum relationship between SPD and schizophrenia. The authors conclude that the evidence for including BD in this cluster is limited and including SPD presents conceptual problems as it requires the absence of psychotic symptoms.

Goldberg et al. (pp. 2043–2059) assessed the validity of the emotional cluster. The authors note that the defining characteristic of this cluster is the presence of negative affect. There are additional similarities that the authors argue support the feasibility of an emotional cluster, and strong intra-cluster co-morbidity may reflect the operation of common risk factors. The authors conclude that emotional disorders meet many of the 11 criteria for determining the validity of clusters.

Krueger & South (pp. 2061–2070) assessed the validity of the externalizing cluster. The authors note that this cluster distinguishes itself by the central place of disinhibitory personality in the disorders grouped within it, including substance dependence, antisocial personality disorder and conduct disorders. The authors further note that shared biomarkers, co-morbidity and course offer additional evidence for the validity of the externalizing cluster.

In the final paper of this group, Goldberg et al. (pp. 2071–2081) further considered the position of bipolar disorder (BD), given that it could be with psychoses, emotional disorders or a separate cluster. In reviewing the literature on BD, unipolar depression and schizophrenia, the authors found that there were notable differences on most of the 11 validating criteria between BD and both unipolar depression and schizophrenia. The authors conclude that there are problems with grouping BD with the psychoses and with unipolar depression. An alternative is for BD to be in a cluster of its own.

Commentaries

In the first of three commentaries on the proposed meta-structure for DSM-V and ICD-11, Wittchen et al. (pp. 2083–2089) raise a number of questions. These include concerns about whether some of the clusters represent a return to earlier simpler classifications (e.g. neuroses, psychoses) and whether, in using a top-down approach to decide on initial clusters, the various authors tend to focus on similarities between disorders rather than differences. Related to this, the authors note that some of the evidence presented may be overstated, notably claims about causal factors. In light of these concerns, the authors conclude that the meta-structure should be considered a useful exploration rather than a solid basis for revisions to DSM and ICD.

In the second commentary, First (pp. 2091–2097) begins by noting that the costs and benefits of the proposed meta-structure need to carefully weighed. The author argues that changes to the current system will carry costs and that the evidence for the proposed clusters is, as yet, limited. In addition, First notes that, in the proposed meta-structure, the largest cluster is in fact that for disorders ‘not yet assigned’ which undermines claims that the proposed changes would result in a more parsimonious classificatory system. Consequently, First concludes that the authors of the proposed meta-structure have failed to establish that it is possible at this point to regroup disorders along the lines presented.

In the final commentary, Jablensky (pp. 2099–2103) raises a number of further points regarding the proposed meta-structure, including the limitation that the literature reviews used to assess the validity of the proposed clusters were not systematic, with the consequence that they ultimately reflect expert opinion based on selective reviews. In addition, Jablensky questions the degree to which the proposed classificatory system would be useful in clinical practice, arguing that assigning individuals to clusters would not be straightforward. Jablensky concludes that the evidence does not yet support a new lumping of disorders into the proposed clusters.