Can doctors predict patients' abbreviated mental test scores

Objectives: the abbreviated mental test is widely used in the assessment of cognitive impairment in elderly patients. However, many doctors do not administer the full 10 questions, preferring to estimate the patient's score instead. We have studied the accuracy of doctors in predicting patients' abbreviated mental test scores. Methods: we assessed 102 patients in the geriatric unit. We asked doctors to predict the patient's abbreviated mental test during the admission interview. A true abbreviated mental test was then recorded. Results: mean age was 80.9 years with a male:female ratio of 27:74. The mean predicted abbreviated mental test score was 6.57 (SD 2.9); the mean actual abbreviated mental test score being 6.36 (SD 3.2). Comparing the two groups, abbreviated mental test scores were predicted most accurately at the extremes and correlation between the two groups of scores was high (P<0.001 Spearman test). Kappa statistics revealed moderate agreement between the two groups, (0.56, 95% CI 0.48-0.63). A predicted score of 5/10 showed the greatest spread of true abbreviated mental test scores (0-10, mean 4.5). However in total, only 31% of the predicted abbreviated mental test scores were accurate, with 42% being incorrect by >1. Using the accepted cut-off of <7/10, this revealed that 13% were underdiagnosed and 19% were overdiagnosed as being cognitively impaired. Conclusions: clinicians are poor at predicting abbreviated mental tests in the midrange but are more accurate at predicting lower and higher scores. This descriptive study reinforces the importance of using an objective assessment of cognitive impairment rather than clinicians estimating its presence or absence.