Abstract
The dimensional models underpinning both recent DSM and ICD systems assume that depression is an entity varying dimensionally, most commonly by severity but allowing other dimensional parameters (e.g. duration, recurrence and persistence). Our Black Dog Institute model [1] assumes that there are two categorical types of depression (i.e. psychotic and melancholic depression) definable by the presence of two specific clinical features and that there is a third heterogeneous residual class of non-melancholic depressive disorders. Our model assumes that the depressive mood disorder is ubiquitous across the three classes, albeit (in general terms) increasing in severity from non-melancholic to melancholic and, in turn, to psychotic depression. Melancholic depression is distinguished by a specific additional component (i.e. observable psychomotor disturbance, or PMD) which (again in general terms) is even more severe in psychotic depression (but here, the class specifying clinical feature is the presence of delusions and/or hallucinations). To reduce confounding, our model assumes that individuals must have a primary diagnosis of depression and that they are at, or near to, the nadir of that particular episode.
Ours is akin to Fould's ‘inclusive non-reflexive’ hierarchical model [2], but perhaps better viewed as a recruitment model, in that we argue that the depressed patients occupying the higher classes must have the features of the lower class (i.e. those with psychotic depression must also have PMD and a depressed mood and those with melancholia must have a depressed mood). The ‘recruitment’ model effectively posits that certain processes underpin the mood disorder and that additional processes underpin the PMD and psychotic feature components, rather than view the latter as determined merely by any greater severity of the mood component drivers.
Jim Greenwood is perhaps arguing for a strong test of any model (i.e. refutability of the hypothesis), although movement in and out of the disorder episode does not strike me as central and could be a major study confounder. Representative samples would be needed for testing rather than mere selection of an ‘exception to the rule’. Any such study might best involve determining if ‘psychotic depression’ exists without PMD and mood disturbance. I would be suspicious that such patients would be more likely to have schizophrenia or a psychotic condition other than psychotic depression – but this is testable. It would also require determining if melancholia ever existed without PMD. This is less readily testable as there is ‘Melancholia’ and ‘melancholia’. While we have argued that PMD is a ‘necessary and sufficient’ feature of melancholia [3], it is only one of eight DSM-IV criteria features, with those decision rules allowing melancholia to be diagnosed in the absence of PMD.