Can you really make that assumption, based on the evidence? Let’s explore it in detail.
ARGUMENT: Bilingualism delays age at onset of dementia, independent of education and immigration status.
The purpose of the study was to determine the association between bilingualism and age at onset of dementia and its subtypes, taking into account potential confounding factors.
Case records of 648 patients with dementia (391 of them bilingual) diagnosed in a specialist clinic were reviewed. The age at onset of first symptoms was compared between monolingual and bilingual groups. The influence of number of languages spoken, education, occupation, and other potentially interacting variables was examined.
Overall, bilingual patients developed dementia 4.5 years later than the monolingual ones. A significant difference in age at onset was found across Alzheimer disease dementia as well as frontotemporal dementia and vascular dementia, and was also observed in illiterate patients. There was no additional benefit to speaking more than 2 languages. The bilingual effect on age at dementia onset was shown independently of other potential confounding factors such as education, sex, occupation, and urban vs rural dwelling of subjects.
This is the largest study so far documenting a delayed onset of dementia in bilingual patients and the first one to show it separately in different dementia subtypes. It is the first study reporting a bilingual advantage in those who are illiterate, suggesting that education is not a sufficient explanation for the observed difference. The findings are interpreted in the context of the bilingual advantages in attention and executive functions.
Full study can be accessed here; unfortunately it has not been made publicly available, free of charge. Please do check if your institution has access to the journal. It can be accessed here as well (which is, at the time of writing, free of charge)
[Copyright 2013 American Academy of Neurology, Alladi et al.]
COUNTER: You cannot say that, unless you control for the age distribution in your group.
From their study of 648 patients with dementia in a clinic in India, Alladi et al. concluded that bilingualism leads to a delay in onset age of dementia compared with monolingualism.1 The data are not sufficient to draw this conclusion. Age at onset studies conducted in a single disease group cannot indicate associations with risk factors because the age at onset depends on the age distributions of the groups from which the participants are selected in the source population.
If monolingual persons die at a younger age on average than their higher-educated bilingual counterparts, then the mean age of monolingual people in the source population will be lower than that of bilingual people. This difference will be reflected in the mean age at onset of any disease of later life, including dementing illnesses. To establish that the finding reflects cognitive reserve and is not an artifact of differing age distributions of monolingual and bilingual people in the source population, the age distribution of the source population by monolingualism/bilingualism needs to be determined.
In this study, monolingual patients had a lower education level and greater illiteracy, and they more frequently lived in a rural environment. These factors are known to be related to lower life expectancy. Adjusting for these factors in models restricted to cases cannot address the issue of differences in life expectancy in these groups in the source population.
James A. Mortimer
Dr. Mortimer raises the important issue of confounding variables, which are relevant to all observational studies. Our study controlled for these variables1 to a higher degree than others. We succeeded in eliminating the immigration confound. We also examined illiterate monolingual and bilingual participants separately and found an even larger difference than among literate participants. Following Dr. Mortimer’s comments, we conducted an additional analysis, examining rural participants (n = 149) separately, and still found a difference between monolingual and bilingual participants (56.2 [10.9] vs 60.7 [9.6] years, p < 0.01).
Our study was not a single disease study; our cohort comprised different types of dementia with different etiologies, symptomatologies, and, of importance, ages at presentation. If the differences in the age at onset were due to lower life expectancy among monolingual participants, we would expect to find the largest effect in those dementia types that present late. We found the opposite. The largest difference was found in frontotemporal dementia, which presents earlier than other dementias.2 In Alzheimer disease, presenting almost a decade later, the difference was smaller, and in dementia with Lewy bodies and mixed dementia—the highest age at presentation—the difference was the smallest. Future studies should further minimize potential confounding variables and this will better clarify these associations.
Question for discussion: Do you think that the author’s answer is sufficient? Can you conclude that the hypothesis is supported by the evidence?