A reevaluation of the duration of survival after the onset of dementia. (Abstracts of Current Literature).

Wolfson C, Wolfson DB, Asgharian M, et al (Department of Epidemiology and Biostatistics, Department of Mathematics and Statistics, and Division of Geriatric Medicine, McGill University, Montreal, Quebec; Department of Mathematics and Statistics, York University, Toronto, Ontario; Department of Community and Family Medicine, Duke University, Durham, NC; Division of Geriatric Medicine, Dalhousie University, Halifax, Nova Scotia; Department of Medicine, University of Calgary, Calgary, Alberta), N Engl J Med. 2001;344:1111-1116.

The purpose of this study was to re-evaluate the duration of survival of people after the onset of dementia. This study also took into account those patients with rapidly progressive illness who died before they could be included in a study. The authors refer to this phenomenon as "length of bias." The authors hypothesize that the previous calculations of the median survival rate after onset of dementia of 5 to 9.3 years have been overestimated. This prospective study estimated the duration of survival from the onset of dementia while adjusting for length of bias.

The sample for this study was identified from the Canadian Study of Health and Aging, which consisted of 10,263 randomly selected Canadians 65 years of age or older who agreed to be interviewed. The sample was screened for cognitive impairment with the Modified Mini-Mental State Examination (MMSE). Cognitive impairment was suspected in those participants with a score of 78 out of 100 or lower, and these participants were referred for a 3-part clinical examination. Part 1 was carried out by a nurse who performed the MMSE, screened the participant's hearing and sight, recorded vital signs, and recorded pertinent family and personal history. The second part consisted of a physical and neurological examination carried out by a physician. Part 3 of the examination was supervised by a neuropsychologist and consisted of a battery of tests administered to all subjects who scored 50 or higher on the second MMSE.

The physician and the neuropsychologist, both of whom were unaware of the scores on the MMSE, arrived at a consensus diagnosis of "dementia," "cognitive impairment but no dementia," or "no cognitive loss." Those subjects with a cognitive loss (n=1,132) were further classified as "probable Alzheimer disease," "possible Alzheimer disease," or "vascular dementia." Of the 1,132 possible subjects with cognitive loss, 185 were excluded because the date of onset could not be ascertained. One hundred seventy-five participants were also excluded because they were classified with another form of dementia. (This group (contained 51 subjects who were also missing dates of onset.) Two participants who died more than 50 years after the onset of symptoms were considered outliers and dropped from the study. A total of 821 participants were followed for 5 years, and, at that point, the participants received follow-up examinations or, in cases in which the participants died, a family member was interviewed to determine the cause and date of death.

For the entire sample, the unadjusted median survival was 6.60 years (95% confidence interval [CI]=6.22, 7.09). After adjusting for length of bias, the estimated median survival was 3.33 years (95% CI=2.66, 4.00), much lower than reported in previous studies. There was no difference in survival when the diagnostic groups were compared. The participants' education had no effect on survival. The authors also determined that being female and an onset of dementia at a younger age were predictors of longer survival.

According to the authors, the unadjusted overall survival median of 6.60 years was within the range reported in previous studies. They argued that the adjusted median of 3.33 (95% CI=2.66, 4.00) demonstrates that the difference is not due to sampling error because the adjusted median is comparable to other serious diseases in older people (eg, other studies report a survival for congestive heart failure of 3-5 years).

The authors pointed out that 134 subjects were missing data on the amount of education they received, so they were excluded from the predictive model because it included this variable. On subsequent analysis, the authors reported that the group with missing education data had more severe cognitive impairment than the remainder of the cohort and had a shorter survival, a finding that the authors felt could explain some part of the discrepancies between some of the results.

Joseph A Lucca, PT, PhD
University of Delaware
Newark, Del