Article Text

Download PDFPDF

Performance on the praxis subscale of the CAMCOG was associated with survival in patients with Alzheimer's disease
  1. Rupert McShane, MRCPsych
  1. Department of Psychiatry of Old Age Churchill Hospital Oxford, UK

Statistics from

OpenUrlCrossRefPubMedWeb of Science

Question How do scores on specific items of the cognitive subscales of the Cambridge Examination of Mental Disorders (CAMCOG) neuropsychological test relate to survival in patients with early Alzheimer's disease (AD)?


Inception cohort with 5.7 years follow up.


Outpatient memory clinic at the University of Amsterdam, the Netherlands.


163 patients diagnosed with probable or possible AD according to NINCDS-ADRDA criteria. Survival data were available for 157 patients (96%, mean age 79 y, 59% women).

Assessment of prognostic factors

At baseline patients were evaluated using CAMCOG, the Cambridge Examination of Mental Disorders of the Elderly (CAMDEX-N), and the Mini-Mental State Examination (MMSE). Demographic data were also collected.

Main outcome measure


Main results

80 patients (51%) died during follow up. Mean survival was 40 months. Lower performance on the praxis subscale was related to decreased survival (hazard ratio [HR] 0.77, 95% CI 0.70 to 0.84, p<0.001). After adjustment for age and sex and when all CAMCOG subscales were entered into the model, praxis was still the only subscale associated with survival (HR 0.82, CI 0.72 to 0.93, p<0.01). After adjustment for age, sex, education, total CAMCOG score, and diagnosis of probable or possible AD and when all 8 of the praxis subscales were entered into the model, only 2 of the praxis subscales were associated with survival: drawing a spiral (HR 0.56, CI 0.33 to 0.96, p<0.05) and drawing a 3 dimensional house (HR 0.25, CI 0.12 to 0.55, p<0.001). When drawing a spiral and a 3 dimensional house were combined into a new copying score, those with the lowest score (poorest copying ability) had a median survival of 32 months, those with a mid-range score had a median survival of 53 months, and those with the highest score (best copying ability) had a median survival ≥62 months.


In patients with Alzheimer's disease, performance on the praxis subscale of the CAMCOG was associated with survival independent of age and sex.


Caregivers of patients with AD occasionally ask “How much longer is he going to live?” The question is rarely answered directly, but a ball park figure can be useful to help both patient and carer plan for the future. If this study by Claus et al is taken at face value, the answer to their question might be “about 3 years, on average” if the patient cannot draw a spiral, and “at least 5 years” if they can draw a 3 dimensional house.

How could such a simple test predict survival? The authors suggest that the constructional test is not simply a test for poor cognitive function or severity of dementia and speculate that it is parietal lobe involvement which impairs performance and increases mortality. But the parietal lobe does not subserve functions critical for life. The numbers of people who die in accidents because of topographical disorientation can hardly account for the effect. Furthermore, the authors report in other studies based on the same series of patients, that cognitive function is a predictor of survival. Another possibility lies in the poor performance of the McKhann criteria for AD in excluding cases of dementia with Lewy bodies (LB). Good evidence exists showing relatively impaired constructional ability when LB pathology is present in dementia. Most studies find that LB pathology is slightly more common in men and that those with LB pathology dementia die sooner.

A model for predicting prognosis of AD has been prospectively validated in only 1 study and the authors from Columbia University have posted their sophisticated, yet easy to use, model on the internet.1 When the patient's sex, MMSE score, age, age of onset (before or after 65 y), duration of illness, presence of psychosis and extrapyramidal signs are entered, information is given in the following form: 25% with these parameters died within x months, 50% died within y months, and 75% died within z months.

The hypothesis presented in this study needs validation before being applied clinically. Until then, the replicated predictors of decline—parkinsonism, psychosis, age of onset, male sex, age and dementia severity—are probably more valuable prognostic indicators than constructional ability.

View Abstract


  • Source of funding: National Committee on Investigative Medicine of the Health Insurance Executive Board of the Netherlands.

  • For correspondence: Dr J J Claus, Department of Neurology, Academic Medical Centre, University of Amsterdam, Meibergdreef 9, 1105 AZ Amsterdam, the Netherlands. Fax +31 35 688 7080.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.