Depression And The Threshold Problem

Another important empirical aspect of affective disorders is the distribution of symptoms in the general population. Many people have a few symptoms, while few people have many. This means that decisions have to be made about the threshold below which no disorder should be identified. People who have few symptoms may still be above this threshold if some of their symptoms are particularly discriminating, but, in general, the threshold is defined by the number of symptoms. There is always a tendency in medicine to move the threshold down, particularly as a sizeable proportion of the people with mental symptoms who are seen by primary-care physicians fall below the thresholds of DSM-IV or ICD-10. However, others in the medical profession have serious reservations about what they regard as medical imperialism, the medicalization of normal human experience (Double, 2002).

In response to the threshold problem, there has been a burgeoning literature recently relating to subthreshold, subclinical, minor, and brief recurrent affective disorder (Schotte & Cooper, 1999). The tendency to extend the threshold downwards is apparent in the establishment of the category of dysthymia, referred to above, a depressive condition characterized only by its mildness (that is, a lack of symptoms) and its chronicity. The category has, nevertheless, become a study it its own right: it has clear links with major depression, presumably because it is relatively easy for someone who already has some depressive symptoms to acquire some more and meet criteria for the more severe disorder. It is also associated with psychosocial distress, both recent and distant. Some authors have gone so far as to suggest that it reflects abnormalities of neuroendocrine and neurotransmitter function (Griffiths et al., 2000).

The imposition of a threshold on an apparent continuum lacks some of its arbitrariness if it is possible to demonstrate a naturally occurring 'step-change' in the distribution. Thus, while the distribution of IQ is largely continuous, there is a clear excess of subjects at the bottom of the continuum who are characterized by a distinct and identifiable pathology (Penrose, 1963). Many have argued that no such distinction exists in affective symptoms (Goldberg, 2000; Tyrer, 1985). While it might be possible to create a threshold that represented a step-change in social disability (Hurry et al., 1983), the evidence does, overall, suggest that affective symptoms are distributed more like blood pressure than IQ. Melzer and his colleagues (2002) have recently used symptom data from the British National Survey of Psychiatric Morbidity to test the smoothness of the distribution. A single exponential curve provided the best fit for the whole population, but there were floor effects that produced deviations at symptom counts of 0-3. Truncation of the data to take account of this provided an excellent fit (Figure 1.1). This was not affected by selecting for analysis subgroups characterized by especially high or low prevalence.

It can be concluded from this discussion that the epidemiological literature on depressive disorder is likely to be a mess. We have disorders that are identified as classes imposed on what is empirically a continuum. This is made worse by the fact that the classificatory schemes are changed at regular intervals. Moreover, two major schemes exist side-by-side. Added to this is the issue of how the symptoms of common mental disorders can be elicited,


Proportion of population




CIS-R score

CIS-R score

Figure 1.1 Proportion of population by truncated range of CIS-R scores, and fitted exponential curve. Reproduced from Melzer, D. et al. (2002). Common mental disorder symptom counts in populations: Are there distinct case groups above epidemiological cut-offs? Psychological Medicine, 32, 1195-1201. Reproduced by permission of Cambridge University Press identified, and used, in order to decide whether, together, they can be said to constitute a case.

Case identification is the basis of epidemiology. The process of diagnosis involves allocating symptom patterns to a diagnostic class according to given rules. In recent years, these rules have been set out explicitly in the diagnostic criteria for research (DCRs) attached to specific classifications, such as DSM-III-R, DSM-IV, and ICD-10, so precisely that it is possible to incorporate them into computer algorithms such as CATEGO (Wing et al., 1990) and OPCRIT (McGuffin et al., 1991).

Once the presence of symptoms has been established, the information can be entered into one of these computer programs in order to provide a diagnostic classification. Human idiosyncrasy can be reduced to an absolute minimum in this process. However, researchers must still decide how carefully the underlying symptoms should be identified. The choices include unstructured clinical assessment, responses to questionnaires, and semi-structured research interviews.

The first option, unstructured clinical judgement, introduces variability into the process of case allocation, since researchers are relying merely on their devotion to a common educational tradition. This situation is even worse when the judgements of others (for

Beat Depression Now

Beat Depression Now

Let me be up front. My intention is to sell you something. Normally, it's not wise to come out and say that. However, I can do so because I have such an incredible deal for you that you'd be crazy to pass on it.

Get My Free Ebook

Post a comment