The clock-drawing test: a critique.

Let me begin with a declaration of disinterest. The esteemed editor of International Psychogeriatrics commissioned this article after reading my letter to Another Journal (Philpot, 2003) in which I called for a moratorium on all clock-drawing test (CDT) research. It seemed to me that an awful lot of effort was being put into a fairly trivial issue and that researchers’ time might be better spent. That much effort is being expended cannot be denied. Using the search words ‘clock drawing’, MEDLINE currently lists 177 publications, 91 since the turn of the century. The first reference to this specific form of drawing test comes in 1983 (Goodglass and Kaplan), and the first paper on validation is that of Shulman et al. (1986). The CDT has emerged as the front runner from a range of drawing tests that have been used to investigate cognition over the last 40 years. When I trained at the Maudsley Hospital in the early 1980s we used the ‘draw-a-cat’ test and used what is now known as a ‘naı̈ve’, informal scoring method to judge ‘normality’ or ‘abnormality’. Cube-drawing used to be popular but is strongly influenced by educational and cultural background (Rosselli and Ardila, 2003). The problem of three-dimensionality and perspective also applies to the house-drawing test (Moore and Wyke, 1984), but is the specific focus of the ‘draw-a-tree-linedavenue’ test (Rennert, 1971). The House-Person-Tree test can be used to assess intellectual function, as well as having a projective element (Fukunishi et al., 1997). The properties of any screening test should not only include ease of administration and acceptability but ease of scoring and, if used sensibly as a first stage in disease identification, there should be a high sensitivity and positive predictive value. Specificity and negative predictive value are less important in this respect, unless the screening method is itself lengthy, aversive or expensive, none of which apply to any of the forms of the CDT. It seems to this reviewer that, while the CDT is often proposed as a means of identifying patients with early stages of dementia who might benefit from specific treatment, it has largely been validated in groups of patients with established diagnoses. Or worse still, validated in patients whose ‘dementia’ is only established by scores on other screening tests. Ken Shulman (2000) has written an excellent and balanced review of the CDT in which he emphasized that it should be used in conjunction with other broader tests of intellectual function, such as the Mini-mental State Examination (MMSE), the verbal fluency test and informant reports. My quibble here would