The question posed by Schrag et al 1(this issue, pp 308–312) is fundamental, but can it be answered? Quality of life belongs to a family of terms which has an improbably wide range of applications, from “quality” as an object of philosophical contemplation2 to “QoL” as an economic variable.3 It must be borne in mind that measures such as the PDQ-39 and the SF-36 differ considerably in content. More detail on the parallelism between the PDQ-39 and two other measures would be helpful but the data seem to support the assumption, which is gaining ground in the literature, that such measures reflect a single underlying dimension. Scales specific to Parkinson's disease and to inflammatory bowel disease have been found to be congruent, again suggesting that disease specific quality of life is a unitary construct.4 However, we must not lose sight of the fact that at the deepest level, quality of life is inherently unquantifiable.

Philosophical considerations aside, there is practical merit in the more precise question: “What are the associations betweenthis particular measure and other variables?” We need to understand which aspects of Parkinson's disease weigh most heavily on patients. Although the authors recognise that associations do not indicate causality their bias is revealed when they write of the “influence” of factors such as mood on quality of life. The converse is equally likely. The two dimensions, mood and quality of life, are logically related, and the extent to which the Beck depression inventory correlates with the PDQ-39 tells us as much about the nature of the two measures as about the patient population under study. Much of the variance was not shared, suggesting that a fraction of the putative quality of life construct did not influence mood (why not?); or that aspects of the measure of mood did not influence quality of life: which sounds paradoxical.

The authors found that severity of disease (or rather severity of physical impairment, as this is what the Hoehn and Yahr scale measures) was more predictive than disability of the PDQ-39 score. The relation between physical impairment and quality of life therefore seems to have been indirect, and not mediated through the obvious effects of impairment on activities of daily life. This is consistent with several other studies showing that the association between disability and quality of life is weak.5 Severity of impairment might have been a marker for diverse experiences which were not individually predictive of PDQ-39 score. It is not clear whether the association between postural instability and PDQ-39 was direct or indirect.

Similar considerations apply to the association between the mini mental state examination (MMSE) score and PDQ-39. We should not accept uncritically the implication that dementia is experienced negatively. It could be an indirect marker of some other negative influence, such as carer stress. One problem which the authors do not consider is that quality of life and also depression are notoriously difficult to measure validly in people with dementia. In this sort of study it is virtually impossible to avoid the influence of carers' opinions on questionnaire responses, especially at the lower end of the MMSE scale, and the authors apparently made no attempt to do so. Carers add a further layer of complexity to the interpretation of these results.