Bilingualism delays age at onset of dementia, independent of education and immigration status

DISCUSSION

This is the largest study so far to examine the impact of bilingualism on dementia and to document a significant delay in the age at onset of dementia symptoms in bilingual patients in comparison to monolingual patients. It is the first study to demonstrate a delay in 3 different types of dementia: not only in AD dementia, but also in FTD and VaD. In contrast to previous studies, the bilingual group was drawn from the same environment as the monolingual one and the results were therefore free from the confounding effect of immigration. The bilingual effect on age at dementia onset was shown independently of other potential confounding factors, such as education, sex, occupation, cardiovascular risk factors, and urban vs rural dwelling, of subjects with dementia. Importantly, for the first time, the bilingual effect was also observed within the illiterate population, demonstrating that bilingualism effects cannot be reduced to differences in education. The observed effects were independent of the specific language spoken. The delay in onset of dementia was almost identical to that reported in previous studies: 4.5 years. Taken together, our results offer strong evidence for the protective effect of bilingualism against dementia in a population radically different from populations studied so far regarding their ethnicity, culture, and patterns of language use.

Interestingly, we could not reproduce the beneficial effect of the number of languages spoken, as advocated in previous studies.² In our cohort, knowing 3 or more languages did not confer an additional advantage over 2 languages. However, this apparent contradiction could be resolved by considering the differences in several individual and interactional factors related to the pattern of language acquisition, proficiency, and use among different countries. The individual factors include language-learning histories, whether the 2 languages were learned in an informal naturalistic context or whether some languages were learned in formal settings, and whether any one of the various languages was dominant or used more often. The interactional factors include the specific context in which languages were used, the linguistic abilities and preferences of people communicating, and degree of relatedness of the languages, which in turn affects translatability and encoding difficulty or ease.¹⁹

Following the influential model of Green²⁰ (1998), language switching has been considered as central in establishing the cognitive advantages of bilingualism. The constant need in a bilingual person to selectively activate one language and suppress the other is thought to lead to a better development of executive functions and attentional tasks with cognitive advantages being best documented in attentional control, inhibition, and conflict resolution.^7,8,21 In places in which an official dominant language coexists with a number of minority languages, it can be reasonably assumed that the amount of language switching between languages is proportional to the number of languages spoken: the more languages people know, the more occasion they will have to switch between them. In the strongly trilingual environment of Hyderabad with Telugu, Dakkhini, and English being used extensively and interchangeably in both formal and informal environments,¹¹ with high levels of code switching and mixing,¹² it could be speculated that those speaking 2 languages have already reached a maximum level of switching and the knowledge of additional languages will not be able to increase it. Such an interpretation would be supported by the view that neural mechanisms underlying cognitive control demands in bilingual communities with high levels of code switching are different from bilingual communities with practice in avoiding language switching or mixing.²²

The effect of attention could also explain the fact that a bilingual advantage was found across different dementia subtypes. Although different forms of dementia have characteristic cognitive profiles, such as episodic memory loss in AD dementia or behavioral impairment in FTD, attentional and executive deficits are common in the early stages of all dementias. Speed of attentional switching and inhibition are found to be impaired in early AD dementia,²³ FTD and VaD are characterized by prominent executive dysfunction,^24,25 whereas DLB is characterized by prominent attentional control deficits.²⁶ The delay in initial symptom onset in bilinguals across dementia subtypes may therefore be related to the protection from decline of attention and executive functions, as observed in normal aging. The fact that we did not find a significant difference between bilingual and monolingual dementia patients with DLB and mixed dementia does not exclude a potential impact of bilingualism, given the fact that there was a trend toward delay and the size of the 2 groups was small and therefore did not allow detection of subtle differences. Further research is needed to address this question.

Several lifetime experiences such as education, occupation, intellectually stimulating activities, physical exercise, and socioeconomic status have been found to delay the age at onset of dementia by improving “cognitive reserve.”^5,27,28 One of the main criticisms directed toward studies claiming a relationship between dementia onset and bilingualism is the effect of potential confounding factors, making it difficult to trace any possible genuine effect of bilingualism. Education has been widely studied, but the results so far have been contradictory: in some studies, education could explain the bilingual advantage, in others it could not.^1,2,29 In our study, although bilinguals were overall more educated than monolinguals, the bilingual effect was shown to be independent from education. Moreover, a significant bilingual delay was demonstrated in the group consisting entirely of illiterate patients without any formal schooling. Similarly, other factors that were unequally distributed between the monolingual and bilingual groups (occupational status, male sex, urban dwelling) did not show a significant interaction with bilingualism. Finally, in contrast to most previous studies, our cohort consisted almost entirely of people who grew up in the Hyderabad area and the state of Andhra Pradesh, minimizing the influence of migration as a confounding variable.

Our study has some limitations. First, age at onset of dementia was determined based on the reports from patients' family members in a clinical setting, and therefore could not be validated. However, because the same criterion was applied to all patients, a systematic bias affecting the results is less likely. Second, similar to most studies examining the relationship between bilingualism and the onset of dementia,^1,–,3 the definition used for bilingualism was a subjective rating. Objective methods of assessing proficiency in languages were not used. Third, a selection bias might have occurred because all of the patients were those reporting at the specialist clinic and not from the community. This is reflected in the age at onset of our cohort, which was nearly a decade less than reported from epidemiologic studies in India,^30,31 although consistent with Indian memory clinic studies.^16,32 Sociocultural protection of elderly and higher levels of comorbidities leading to lower referral rates of elderly, and younger demographic profile due to lower life expectancy,³³ are likely to have accounted for this finding. Finally, in the absence of information on how each bilingual patient acquired all the languages they knew, it was not possible to make comparisons based on language combinations.

The converging evidence from different research groups, countries, and cultures, all pointing to bilingualism as a potential protective factor against dementia, makes it increasingly less likely that the bilingualism effect can be reduced to artifacts and confounds. Further studies are needed to substantiate the claim and elucidate the possible neural mechanisms. There is a need for a prospective study, collecting medical, cognitive, linguistic, and social information, ideally in combination with neuroimaging. The linguistic data to be collected should include more comprehensive information about the language acquisition and proficiency. However, what we consider to be most important is the need for comparative studies in different countries, languages, and cultures. Our study demonstrates that investigating a new population with a different pattern of variables associated with bilingualism can unearth insights that would be difficult to obtain by limiting the research to previously studied populations. We hope that our research will stimulate interest in the interactions of bilingualism, cognition, and dementia across the world.