Differences in use of abbreviated mental test score by geriatricians and psychiatrists

BMJ 1996; 313 doi: (Published 24 August 1996) Cite this as: BMJ 1996;313:465
  1. John Holmes, senior registrar in old age psychiatrya,
  2. Simon Gilbody, tutor in psychiatrya
  1. a Department of Liaison Psychiatry, Leeds General Infirmary, Leeds LS1 3EX
  1. Correspondence to: Dr Holmes.
  • Accepted 3 December 1995

Cognitive impairment is common in elderly people and is associated with increased morbidity and mortality,1 but confusion is often poorly recognised and documented by medical staff.2 3 Thus, routine cognitive screening of elderly patients in hospital has been recommended.4

One validated and widely used screening instrument is the abbreviated mental test score, in which the maximum score is 10 and a score below 7 suggests cognitive impairment.5 It is widely used in clinical and research settings in Britain for detecting and monitoring cognitive impairment and is easily administered and well tolerated by raters and subjects. Inconsistencies in giving and scoring the test will affect its reliability, validity, and sensitivity in detecting change. We noticed differences between colleagues in the use and scoring of this test and investigated how it was applied in routine clinical practice.

Subjects, methods, and results

We contacted 105 doctors who regularly assessed elderly patients; one was not aware of the abbreviated mental test score and was excluded. Of the remaining 104 doctors, 58 were psychiatrists (six consultants, five staff grades, four senior registrars, 19 registrars, and 25 senior house officers) and 46 were geriatricians (11 consultants, three senior registrars, five registrars, and 27 senior house officers). We asked them to list the items they used when giving the test; doctors referred to preprinted questions if they usually used them in their clinical practice.

Only 23 doctors used the 10 items from the original validated version of the abbreviated mental test score. The median number of original questions asked was 8 (interquartile range 6-9). Table 1 shows the frequency with which items were used. There was no significant difference in the total number of original items used between psychiatrists (median 7, range 1-10) and geriatricians (median 8, range 3-10; Mann-Whitney U test P = 0.260) or between senior house officers (median 7.5, range 3-10) and higher grades (median 8, range 1-10; Mann-Whitney U test P = 0.077). One example of a common substitution was the patient's name, an item which had the lowest sensitivity of all items used in the original validation of the test.5

Table 1

Use of original 10 items in abbreviated mental test score by 104 doctors, with suggested scoring instructions5

View this table:

We then asked each of the doctors to score a set of the same responses to the 10 original items; we said the responses to them with additional information such as the correct age and date of birth of the respondent and the time and place of testing. The correct score was 5, but scores ranged from 3.5 to 9 (mean 6.32 (95% confidence interval 6.08 to 6.57)). No significant difference was seen between psychiatrists (mean 6.17 (5.89 to 6.45)) and geriatricians (6.51 (6.08 to 6.94)) or between senior house officers (6.60 (6.24 to 6.95)) and higher grades (6.05 (5.72 to 6.37)). Seventeen doctors obtained the correct score, but only one of them scored each individual item correctly. Forty one doctors scored 7 or more, incorrectly classifying the respondent as being cognitively intact. Items scored incorrectly were counting backwards from 20 to 1 (85 doctors), place (81), recall of an address (79), and time of day (70). The use of half marks by 78 doctors contributed to incorrect scoring.


We studied doctors working in units where the abbreviated mental test score was commonly used and found inconsistencies in items and scoring that are likely to lead to inconsistencies in practice between doctors. Patients are often tested sequentially, but it cannot be assumed that the abbreviated mental test score as used in clinical practice will be sensitive in detecting changes in cognitive state.

Items testing short term memory and orientation in time were often omitted. These are the components of cognitive function most commonly impaired in dementia and delirium. The removal of these items further questions the validity of the abbreviated mental test score as used routinely.

Medical staff need to be trained in using and scoring the abbreviated mental test score. Preprinted lists of the correct items may be helpful and should be readily available when assessing patients.

We thank Allan House for comments on the manuscript.


  • Funding None.

  • Conflict of interest None.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.