Norton MC, Smith KR, Ostbye T. Greater risk of dementia when spouse has dementia?. J Am Geriatr Soc. 2010 July 1;58(1):895.
Recommends
Please login to recommend the paper.
Norton MC, Smith KR, Ostbye T. Greater risk of dementia when spouse has dementia?. J Am Geriatr Soc. 2010 July 1;58(1):895.
Please login to recommend the paper.
Comments
University of South Florida
Greater Risk of Dementia When Spouse has Dementia: A Spurious Finding?
In this paper, Norton and collaborators show a six times higher risk of dementia in individuals after onset of dementia in their spouses. In the accompanying editorial, Vitaliano notes that when considering only the number of incident dementia cases in the group exposed to a spouse with dementia and in the group without this exposure, there is only a 48 percent increase in risk in the group exposed to a spouse with dementia. How, then, do the authors obtain a 640 percent increase in risk using a Cox proportional hazards regression model? One clue is that those exposed to a spouse with dementia were observed an average of 9.2 years (an average of 5.1 years prior to the exposure and 4.1 years after), while those without this exposure were followed for only a mean of 4.5 years. This very large difference in mean observation time is not explained, but may well provide the explanation for the 640 percent increase in risk.
Table 1 shows that among those with a demented spouse, 14.8 percent developed incident dementia, whereas among those without a demented spouse, 10 percent developed incident dementia. The time over which incident dementia occurred in the two groups is not specified. If one uses 9.2 years for the exposed and 4.5 years for the unexposed, then having one’s spouse become demented is associated with a small reduction in risk per year. If instead one takes the 4.1 year post-exposure interval for the exposed and 4.5 years for the unexposed, there is an approximately 50 percent increase in risk per year. However, under the latter assumption, because the two groups had similar baseline mean ages, the cases in the exposed would have occurred when they were 5.1 years older, on average, which could easily account for the 50 percent increase in risk.
Considering the Cox proportional hazards regression, at any time, all individuals whose spouses are not yet demented are being compared with those whose spouses have become demented. When a spouse becomes demented, an individual changes from unexposed to exposed. In Cox regression, individuals are censored (no longer followed) for reasons other than becoming demented. For the unexposed group, this includes death or dropout from the study of either spouse. For the exposed group, this includes death or dropout from the study of the individual whose spouse previously became demented. Because dementia risk increases with age, most of the cases of dementia in the exposed group would occur after several years of follow-up. By that time, the unexposed comparison group would have lost most of its members through censoring. As noted above, the reason for the large differential in censoring in the two groups is not explained. If the reason for censoring the unexposed group had anything to do with having a higher risk of dementia, then the 640 percent increase in risk reported in this paper could be an artifact of a lower risk of dementia in the survivors of the unexposed cohort and have little to do with an increased risk in those exposed to dementia in their spouse.
References:
Vitaliano PP. An ironic tragedy: are spouses of persons with dementia at higher risk for dementia than spouses of persons without dementia?. J Am Geriatr Soc. 2010 May;58(5):976-8. PubMed.
Make a Comment
To make a comment you must login or register.