Impact of mild cognitive impairment on survival in very elderly people: cohort study

BMJ 1997; 315 doi: (Published 25 October 1997) Cite this as: BMJ 1997;315:1053
  1. J Gussekloo, general practitionera,
  2. R G J Westendorp, senior registrarb,
  3. E J Remarque, research fellowa,
  4. A M Lagaay, senior registrara,
  5. T J Heeren, professorc,
  6. D L Knook, professora
  1. a Section of Gerontology and Geriatrics, Department of General Internal Medicine, University Hospital Leiden, P-3-Q PO Box 9600, 2300 RC Leiden, Netherlands
  2. a Clinical Epidemiology Leiden, University Hospital Leiden
  3. a Department of Psychiatry, University of Utrecht, Utrecht
  1. Correspondence to: Dr Gussekloo
  • Accepted 24 June 1997


Severe cognitive impairment is associated with increased mortality, but the impact of mild cognitive impairment on survival remains unclear.1 2 Although there is doubt whether a simple test such as the mini-mental state examination has sufficient discriminatory power to detect mild cognitive impairment in elderly people,3 we determined the impact of borderline scores in this particular examination on survival in very elderly people.

Subjects,methods,and results

As part of the Leiden 85-plus study4 we followed a cohort of 891 subjects (641 women, 250 men) aged 85 years and over (median age 90 (range 85 −103) years) from 1986 onwards. At entry to the study the score on the mini-mental state examination (Dutch version) was assessed by a physician during a home visit. In cooperation with the local government all but two subjects were followed for survival up to 1 October 1996. In all, 790 subjects died. Relative risks of mortality were estimated in a Cox proportional hazards model, which was adjusted for sex and for age at baseline.

During the first year of follow up, the annual mortality risk for subjects with mild cognitive impairment (score 24-27 points, n=226) was twice as high (relative risk 1.8 (95% confidence interval 1.1 to 3.0)) as the annual mortality risk for subjects with a normal cognitive function (score 28-30 points, n=352). This difference in risk remained similar until the seventh year of follow up, after which the annual mortality risk decreased to unity.

The cumulative mortality risk of the subjects with a mild cognitive impairment during the first seven years of follow up was 1.7 (1.4 to 2.0). This risk estimate was similar for men and women and for subjects below and over 90 years of age at baseline. Compared with subjects with a normal cognitive function, the cumulative mortality risk for subjects with a moderate cognitive impairment (score 19-23 points, n=131) was 2.5 (2.0 to 3.1), and for subjects with a severe cognitive impairment (score 0-18 points, n=180) the risk was 2.8 (2.3 to 3.4).

The association of scores in the mini-mental state examination and mortality is further illustrated in the 1 representing the survival probabilities of subjects, calculated from the age of 86 years onwards.


Survival probabilities from age of 86 years onwards for various categories of scores in the mini-mental state examination. Survival probabilities are calculated on data for 889 individuals followed for seven years, using actuarial method allowing individuals to enter survival table at different years of age (left censoring)


In contrast with general belief, borderline scores in the mini-mental state examination cannot be considered to be normal and are associated with a significant decreased survival. Subjects with mild cognitive impairment may further deteriorate in cognitive function,5 which is again associated with a lower survival.

It is not likely that all observed excess mortality is due to an effect of mild cognitive impairment. It may well be that milder cognitive impairments are associated with physical illness and disabilities, which could by themselves account for the decreased survival. A typical example is that atherosclerosis underlies decline of cognition, as well as cardiovascular disease. It is therefore difficult to determine which part of the observed mortality is due to atherosclerosis and which part to the mild impaired cognition.

Nevertheless borderline scores in the mini-mental state examination discriminate subjects with a higher risk of mortality. In practice, this easy to administer questionnaire seems to be useful as a screening instrument for mild cognitive impairment, and its scores may act as an important predictor of survival in very elderly people.


Funding: This study was partly funded by NIH (grant 5, RO 1 AG 06354) and by the Ministry of Health, Welfare and Sports.

Conflict of interest: None.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
View Abstract