Relation of Literacy and Music Literacy to Dementia in Older Black and White Brazilians 1

INTRODUCTION

Latin America and the Caribbean currently have 56 million persons that are aged 65 years or older [1]. Brazil is the largest country in Latin America and ranks as the 6th country with the largest population of 65 years and older adults, more than 19 million, just behind China, India, United States, Japan, and Russia [1]. The sociological structure of race in Brazil is quite unique. Brazil has a specific category of “mixed” in the Census (IBGE –The Brazilian Institute of Geography and Statistics). Brazilians self-identify as Black, White, or Mixed (termed as “Pardos(as)”). The first author and two co-authors are all native Brazilians. For a variety of historical reasons, in Brazil, interracial marriages have long been and continue to be quite common, and those self-identifying as Mixed are quite frequent in the population. Brazil is also one of the most diverse countries in Latin America. The Brazilian Census uses a question on self-identified race that includes, among others, the categories White, Black, and Mixed. A total of 47%of the Brazilians self-identify as Mixed, 44%self-identify as Whites, and 9%self-identify as Blacks [2], making Brazil a rich environment to study racial differences.

Literacy is one of the cognitive abilities that may provide cognitive reserve and thus higher resilience to decline despite dementia-related neuropathology [3]. Literacy can be acquired from both formal and informal education. A large study on incident dementia including participants from Cuba, Dominican Republic, Venezuela, Peru, Mexico, and China found an overall additional independent effect of literacy, controlling for educational level [4]. Studies of Brazilians have found an association between illiteracy and a higher risk of dementia [5–7]. None of these studies with Brazilians have examined race differences in the effect of literacy on dementia, although this has been studied in other populations [8].

In addition to the more classic literacy (reading, writing, and numeracy), music literacy is an enrichment often available in higher-resourced schooling systems. Music instruction or playing instruments has been shown to be associated with lower odds of cognitive impairment and dementia, higher cognition, and more neuroplasticity [9–14]. Different types of literacy have been studied in the context of cognitive decline [15], but to our knowledge, this is the first study to examine music literacy and dementia in Brazilians. In fact, the majority of studies on music and dementia overall are from Asia, Europe, and the United States [16]. With great racial diversity and the culture of music among all strata of society (e.g. percussion instruments in the Samba Schools and string instruments in Pagode, a subgenre of Samba), the study of music literacy in Brazil can provide a unique contribution to the knowledge of the role of music on protection against dementia prevention among individuals of different races and with lower schooling.

In the present study, we examined the association of different forms of literacy including music literacy and dementia in a large sample of Black, Mixed, and White Brazilians with a broad spectrum of formal education enrolled at autopsy by their legal representative in the Pathology, Alzheimer’s, and Related Dementias Study (PARDoS). We hypothesized that higher literacy would be associated with lower odds of dementia, independent of years of education and other confounders. We also examined racial differences in these associations. We hypothesized the association is dependent on race after important demographics are considered.

MATERIALS AND METHODS

RESULTS

Decedents were aged 65 to 110 years at death. About one-third were Black, half were women, and one-third had dementia. Most of the decedents belonged to social class C (3.4%in social class A, 35%in social class B, 55.1%in social class C, and 6.5%below social class C). Most of the informants were family members, primarily the sons/daughters of the decedents (71.8%) followed by grandchildren (9.1%), siblings (4.9%), and spouses (4.5%). Table 1 shows the key demographic characteristics of the decedents and informants. Education and sex distribution of the informants of decedents both with and without dementia were similar (both with 11 years of education and 53%to 55%of men). Informants of decedents with dementia were on average three years older (51.39, 95%CI = 50.39, 52.39 versus 48.45, 95%CI = 47.75, 49.15 years old) than informants of decedents without dementia.

Table 2 shows a comparison of Black with White decedents on education and literacy variables. The racial groups had the same ranges of literacy and education (range of verbal literacy: 0 to 1; numeracy 0 to 1 and education 0 to 25 years), and sim-ilar ranges of age at death (Blacks: 65 to 110 years at death, Whites: 65 to 105 years at death). Whites, however, had higher median education and literacy (Table 2). Verbal literacy was higher for Whites, even after controlling for years of education (β_adjusted = –0.404, SE = 0.101, p≤0.001) and there was a significant interaction between education and race (β_interaction = 0.111, SE = 0.034, p = 0.001) indicating that the race difference in literacy was higher among individuals with fewer years of education. That is, Blacks had less literacy (a negative effect) but the difference in literacy between Blacks and Whites with few years of education was larger than between Blacks and Whites with more education (a positive interaction effect). Results remained significant when socioeconomic status was added to the model. Men had more years of education (β_unadjusted = 0.82, SE = 0.09, p≤0.001) and higher literacy (β_unadjusted = 0.59, SE = 0.08, p≤0.001), but literacy did not differ by sex after controlling for years of education (β_adjusted = 0.097, SE = 0.089, p = 0.273). This was consistent across race with similar results found when examining Blacks and Whites separately. The odds of dementia also did not differ between Black and White decedents (OR = 0.978, β= –0.0224, SE = 0.1164, p = 0.8474), controlling for sex and age at death.

A total of 309 decedents had zero years of education (51%Black). As expected, they had zero median verbal literacy and numeracy and only 1%knew how to read music. However, some literacy was observed, especially in numeracy as 19%could multiply, 12%could calculate percentages, and 8%could do a simple operation with a calculator. Verbal literacy was extremely low but 10%could use a phone book and 8%could read a phrase.

All forms of literacy were associated with lower odds of dementia (Table 3). Participants in the highest quartile of verbal literacy had 15%lower odds of having dementia compared to the highest quartile. Participants in the highest quartile of numeracy had 27%lower odds of having dementia compared to the lowest quartile. Participants that read music had about half the odds of having dementia. The effect of all literacy types on dementia was similar between Blacks and Whites (Supplementary Table 1, p_s > 0.237).

In the first set of sensitivity analyses, we reran the three models with each literacy type, adjusting age-at-death and sex and, in addition, adjusting for race, socioeconomic status, and education. All literacy remained associated with lower odds of dementia (all p≤0.010). In the second set of sensitivity analyses, we only examined a subset of individuals in social class C that had 4 years of educations. The median literacy and numeracy were respectively 0.80 (IQR = 0.60 to 0.90) and 0.67 (IQR = 0.34 to 1.00), and 6%read music. We reran the three models with each literacy type adjusting for age-at-death and sex. Literacy, numeracy, and music literacy remained associated with lower odds of dementia (all p≤0.034).

In a secondary analysis, we examined the effect of playing an instrument without reading music on the odds of dementia. We used a variable with three categories: reading music, playing an instrument without reading music, and not reading music or playing an instrument (the reference). The effect of reading music remained the same (OR = 0.48, 95%CI = 0.27, 0.85), while the effect of only playing an instrument was smaller and not significant (OR = 0.96, 95%CI = 0.56, 1.66, p = 0.887).

DISCUSSION

In this study with older deceased Brazilians with a wide spectrum of education, we found that multiple types of literacy, including verbal literacy, numeracy, and music literacy, were associated with lower odds of dementia. This adds to a growing literature suggesting that all forms of literacy are protective against dementia. We also show that the effect of literacy on dementia is similar across races. In sensitivity analyses, we showed that the effect was also significant in a group with homogeneous education (all with 4 years of education) and socioeconomic status (all social class C).

Education and literacy (including music instruction) are associated with a higher level of cognition [3, 23]. Education and literacy are thought to build cognitive reserve, which increases the ability to tolerate aging-related changes in the brain without developing clinical symptoms [24]. That can happen through both protective and compensatory mechanisms in a dynamic interaction of the brain and environmental stimuli. Imaging studies suggest that education changes the brain morphology and is associated with regional cortical thickness [25], brain size [26], and microstructural changes of the hippocampus [27]. There is, however, a threshold hypothesis supported by evidence from clinical and biological studies that indicate that there is initial protection against the clinical manifestation of dementia but after brain pathology reaches a critical level there is a more rapid cognitive decline [28], although these mechanisms are not fully understood.

We also show that the effect of all three types of literacy on dementia did not differ by race. To our knowledge, racial differences in the effect of literacy on dementia have not been examined in older Brazilians. The two racial groups had similar benefits from literacy. However, since all types of literacy, except music literacy, were lower in Blacks, this group may benefit more from intervention on literacy. A recent literature review including studies of children with low socioeconomic status suggested that programs that integrate both linguistic and metalinguistic skills are more effective [29]. It also mentions initiatives such as the Venezuelan program “Mochila Mágica” (or the “Magic Backpack”) that lend books and train teachers and parents to support reading at home.

Most of the Brazilian population attending school are enrolled in the public system: 83%of elementary students and 88%of secondary students. Brazil also has free public early child education/daycare for poor families. Although in recent decades inequality has been declining in Brazil [30], high levels of inequality persist and are reflected in the education system. While an upper-middle-class child will have access to education similar to the top education of any developed country [31], the poor struggle to get a few years at an under-resourced primary school. Given the educational inequalities, one year of a private school located in a prosperous neighborhood is not equivalent to one year in a public school located in the poor outskirts of large cities or in remote rural areas. A study examined admission test scores for a federal university in the Northeast region (Universidade Federal de Pernambuco) and found that public school students’ scores were on average 4 to 17%lower than the scores of private school students [32]. On top of the educational inequalities, other factors such as nutrition and childhood stress can also contribute to the literacy inequalities. Although these inequalities in Brazil transcend race, Black and Mixed-race individuals are disproportionally affected by them. We found racial differences in literacy in almost all indicators. Black and Mixed-race individuals were found to have lower verbal literacy, especially among participants with fewer years of education. Although we adjusted for socioeconomic status, our information is limited and does not reflect the resources the person may have had earlier in life.

This study has several strengths. The study includes a large sample size of both Black/Mixed-race and White Brazilians with a wide spectrum of education. It uses informants, mostly the decedent’s children, who knew the deceased for many years. In aging research, informants are commonly used to obtain information on older adults with cognitive impairment and after death. The study also used measures, such as PCAS and CDR, that have been validated as a proxy report and are commonly used with informants in prior research with Brazilians [33]. The study also has limitations. By design, individuals with lower education are over-represented in our study. Although this gives us important variability to study, this is not, however, a population-based study, and generalizability may be limited. In addition, although literacy in the study of dementia is often obtained using reported abilities like reading and writing, literacy was not obtained from proficiency examinations. It is always possible to have some bias and inaccuracies. The decedent may not have shared some information with the family member serving as the informant. For example, they may have learned to play an instrument earlier in life and never mentioned that to their children. Recall bias may also create inaccuracies when data is collected retrospectively [34]. Therefore, the effect may be underestimated as the reference group has more variability. Similarly, the CDR may have a bias, although the window of reference (3 months before death, instead of a lifetime) makes this less likely. Since this study is cross-sectional, we cannot test temporal order. Finally, the question used to assess the role of reading music is just a first look into this research question for this population. Although standard musical notation includes bar lines, treble, and bass clef, other forms of music notations such as guitar/lute-playing and percussion instrumentation include chord designations, string/fingering notations (tablature), or linear representations of strikes. This variability is not captured here. Therefore, further studies can help confirm and extend these findings, including prospectively collected data or studies including brain pathology.