Impacts of Illiteracy on the Risk of Dementia: A Global Health Perspective

INTRODUCTION

The inability to read or write, so called illiteracy, has various harmful effects on the individual as well as on the social level, ranging from daily lives to achieving one’s full potential. In this regard, it has been a focus of human development historically and many governments around the world have put effort to increase the literacy rate of their people.

Nevertheless, literacy rates are reported to vary considerably across the globe. As of 2010, approximately 75% of the worldwide illiterate population was distributed in only 10 countries (in descending order of literacy, these included India, China, Pakistan, Bangladesh, Nigeria, Ethiopia, Egypt, Brazil, Indonesia, and the Democratic Republic of Congo [1]) (Fig. 1). From a geographical perspective, South and West Asia have the lowest regional literacy rate (58.6%), followed by Sub-Saharan Africa (59.7%), and the Arab States (62.7%) [2]. Meanwhile, most highly developed countries, including South Korea, no longer estimate their literacy rate frequently because its figure is already extremely low in the middle-age population, and is expected to continue to decline over the coming decades.

This discrepancy of the literacy rate is of interest to the health policy makers because, as mentioned above, illiteracy itself is a significant burden to a society in many different respects [3]. In economic terms, illiterate people earn salaries, which are 30∼42% smaller in comparison to their literate counterparts. They may also experience serious problems in achieving a vocational education; thus, limiting their future employability and thereby reducing wealth creation opportunities. In social terms, illiteracy can restrict a person’s capacity to access, understand, and apply health related information, which can result in a poorer physical condition with increased demand for medical treatment, and a greater risk of job absenteeism. Overall, this can lead to illiterate people turning to crimes and becoming dependent on social welfare, in both developing, as well as developed countries.

At another point, in neurobiological terms, achieving literacy can enhance cognitive reserve (CR), a hypothetical model which posits that the brain actively attempts to deal with neurodegenerative changes by employing preexisting cognitive processing or by mobilizing compensatory mechanisms [4]. Though it is not mutually exclusive with the CR, brain reserve is another construct referring to ability to tolerate any disease-related pathology in a mature brain without developing clinical manifestations. Both of these two models are thought to be involved in providing the reserve against neurodegenerative changes. Because these reserves vary among individuals, this might be able to explain why there is a discrepancy between pathological severity of dementia and its clinical phenotype.

Abundant studies have proposed that brain volume, synaptic count, and dendritic branching are all possible measures of reserves and are amenable to change by lifetime experiences, including socioeconomic status, occupational attainment, and education [4]. Particularly, educational attainment is one of the most widely used proxy for reserves, for its relationship with the reserve is relatively obvious: by increasing synaptic density in neocortical association area thus delaying symptoms of dementia approximately 4 to 5 years [5]. There are several ways to estimate the educational attainment, and it has been suggested that the degree of literacy can be a superior marker for the CR to the number of years of education because it is a more direct measure of educational achievement[6, 7]. Moreover, literacy provide enriched environment with highly structured and complex activities that give opportunities for self-expression. Together with socioeconomic effects described earlier, this intellectually stimulating environment has been found to act against neurodegenerative insults by enhancing cognitive reserve through neuroplasticity, neurogenesis and locus coeruleus-noradrenergic system [8].

Accordingly, there have been large amounts of evidence that has ascertained a relationship between the literacy and cognitive decline, and/or the development of dementia. Several cross-sectional studies have revealed high literacy rates to be associated with high scores in memory, verbal fluency, and global cognition, as well as other executive skills [9–11]. Furthermore, two longitudinal studies have also demonstrated similar findings (i.e., that as one gets older lower levels of literacy are independently associated with a more rapid decline in language, memory, and executive function [6, 7]). In addition, one recently published article has suggested that limited literacy, especially in combination with the apolipoprotein E ɛ4 negative allele, is a crucial risk factor for developing dementia [12]. While these findings only suggested the correlation between the literacy and the dementia, in his latest article, Nguyen et al. [13] demonstrated that a causalrelationship is at work between the year of schooling and reduced dementia risk. As stated above, the number of years of schooling, though not as direct as the degree of literacy, is a fair measure of the educational attainment which concludes that education lowers dementia probability.

So far, however, few studies have presented accurately to what extent this relationship affects the prevalence of dementia or demonstrated it practically. In this article, we use the population attributable fraction (PAF) to estimate the proportion of dementia cases attributable to illiteracy in South Korea, where its population had maintained a relatively high level of ethnic and linguistic homogeneity up until 1990s [14] among advanced countries listed by Central Intelligence Agency (CIA) [1]. And we applied this method to other continents where the illiteracy rate is extremely high. In addition, we calculate from a global health perspective how much monetary cost can be saved in the future by reducing the illiteracy of a specific age group within populations, at a defined point in time. We hypothesize, that by reducing illiteracy rates, significant proportions of dementia cases will also reduce, with a potential for corresponding increments to savings in dementia carecosts.

MATERIALS AND METHODS

The PAF is defined as “the proportional reduction in average disease risk that would be achieved by eliminating the exposure(s) of interest from a population [15].” To calculate the PAF for dementia caused by illiteracy, we used the following formula as dementia is a sufficiently rare disease for odds ratios (ORs) to be considered an approximation of relative risk [16]. PAF = PPE ( OR - 1 ) PPE ( OR - 1 ) + 1

This formula was originally developed based on a prospective study. However, since illiteracy can only be a causative factor of dementia and not vice versa, we are able to input ORs that have been obtained from cross-sectional studies conducted around the world. In this formula, we also need to define the proportion of a (general) population with this exposure (PPE). PPE of target year n (PPE_n) is expressed in the following way, PPE n = I n / P n where I_n represents the number of illiterate people and P_n represents the total number of people in a (general) population, aged 65 years or over in target year n.

For South Korea, we obtained I_n and P_n data at specific time points from the World Population Ageing: 1950–2050 [17], Population Projections for Korea: 2010–2060 [18], and World Population Prospects: The 2012 Revision [19]. Because the population structures presented in these sources are given by 5-year interval, we gathered data starting from target year 2010 to year 2050 by 5-year intervals. By assuming the “trend scenario” as the literacy projection model [20], future values for specific age groups in the target year were derived from linear extrapolation along a line fitted to the logits of the literacy rates in the observed data. Statistical analyses were conducted using Statistical Package for the Social Sciences (SPSS) for Windows, Version 20.0 (SPSS Inc., Chicago, IL, USA). In addition to South Korea, we also applied the abovementioned methods to several continents where illiterate people are densely populated (e.g., Latin America, Southwest Asia, and Africa). Unfortunately, most prevalence studies conducted in developing countries of these continents did not specifically deal with illiteracy as a risk factor for dementia. We managed to find out several studies with the odds ratio of dementia for illiterates which is either adjusted for multiple independent variables or simply unadjusted (Table 1). Unable to obtain sufficient data to carry out a meta-analysis, we used both the minimum and maximum ORs reported in each continent to calculate respective values for the PAFs. We assumed that the OR of dementia remained consistent throughout time, in any givencontinent.

Furthermore, using expected PAFs and future prevalence rates, we estimated how much the proportion of cases of dementia could be decreased by if illiteracy was reduced in people aged 65 years or under, as of 2015. We assumed that having acquired literacy, it would take 5 years to build up a cognitive reserve which is understood to be protective against developing dementia. As Xu et al. [8] pointed out, very little is known about the timeline between the influence of cognitive reserve and the development of dementia. However, because the CR is increased after acquiring literacy through synapto-dendritic enhancement [5], we were able to find some clues to this matter. Using whole-brain magnetic-resonance imaging, Draganski et al. demonstrated that learning-induced plasticity can alter the brain’s macroscopic structures, not limited to the changes of functional connectivities [21]. And this effect was examined by Carreiras et al. between late-literates, Colombian guerillas, who learned to read as adults after re-integration into mainstream society (mean age; 31.5 years, age range; 21–52 years) and their carefully matched illiterates (mean age; 32.8 years, age range; 20–52 years) [22]. This study revealed that, after at least 5 years of active reading and writing, late-literates had significantly more grey matter than their illiterate counterparts in five brain regions that activated in functional imaging studies of reading (bilateral dorsal occipital areas, left supramarginal and superior temporal areas, angular gyri and posterior middle temporal regions). Moreover, they showed an increased white matter in the splenium of corpus callosum. These structural differences of areas relevant to reading abilities were still valid even in the subgroup analysis of participants matched on a diverse cognitive abilities. Therefore, they are likely the consequences of being able to read, a proxy of the CR. Though it is still not clear whether 5 years are sufficient enough to produce the protective effect against dementia, we assumed that this amount of experience gives late-literates acceptable reserves.

Moreover, we also quantified the potential savings in dementia care costs that could be achieved, if illiteracy was reduced to a manageable extent, as of 2015. To do this, we calculated differences in the PAF before and after reducing illiteracy, and multiplied this by the reported annual dementia care costs from 2010 to 2050 which, in the case of South Korea, was derived from national survey data [23]; for other continents, these figures were obtained from cost estimates of previous studies [24] combined with predicted changes in prevalence [16]. Lastly, we estimated the area under the curve for the annual dementia care costs versus year graphs, which gave us the cumulative amount of money saved by the year 2050.

RESULTS

According to the 2008 Nationwide Survey on Dementia Epidemiology of Korea, the prevalence of dementia and OR of illiteracy for dementia have been reported to be 8.1% and 4.68 (3.57–6.15), respectively [25]. Based on these figures, the PAF of illiteracy for incident dementia was expected to decrease naturally from 15.59% in 2015 to 11.45% in 2020, and 1.62% in 2050, due to a gradual decline in the illiteracy rate of elderly Koreans. The current illiteracy rate of Korean people under 65 years of age is 0.25%. If, as of 2015, we could completely eradicate illiteracy in Korean people under 65 years of age, then the PAF of illiteracy for incident dementia in elderly Korean citizens, aged 65 years or older, would decrease at a much faster rate (i.e., from 11.45% to 9.3% and from 1.62% to 0.03% in 2020 and 2050, respectively). This, in turn, is expected to reduce costs of dementia care; saving approximately 52 billion USD by 2050 in South Korea (Table 2 and Fig. 2).

The current illiteracy rates of people under 65 years of age in Latin America (9.35%), South Asia/The Middle East (33.17%), and Africa (37.93%) are reported to be much higher than in South Korea. The impact of the illiteracy reduction on the PAF of illiteracy for incident dementia in these regions, therefore, has the potential to be substantially larger than in South Korea.

In Latin America, the OR of illiteracy for dementia has been reported to range from 1.83 to 6.84 [26–32]. Assuming the OR of illiteracy for dementia to be at the lower end of this estimate, a 50% reduction of illiteracy in Latin American people under 65 years of age would predict a reduction in the PAF of illiteracy for incident dementia from 14.13% to 12.12% in 2020 and from 7.28% to 3.78% in 2050. This would be expected to save the national economy 70.96 billion USD in dementia care costs by 2050. If we assumed the OR of illiteracy for dementia to be 6.84, then the PAF of illiteracy for incident dementia could decrease much further, from 53.65% to 49.24% in 2020 and from 35.58% to 21.64% in 2050, which would save the national economy 244.07 billion USD by 2050 (Table 3 and Fig. 3).

In South Asia and the Middle East, the OR of illiteracy for dementia has scarcely been reported, with estimates ranging from 1.10 to 3.67. Assuming the OR of illiteracy for dementia to be 1.10, a 50% reduction of illiteracy in South Asian/Middle Eastern people under 65 years of age would predict a reduction in the PAF of illiteracy for incident dementia from 5.32% to 4.28% in 2020 and from 3.41% to 1.73% in 2050. This would be expected to save the national economy 13.09 billion USD in dementia care costs by 2050. If we assumed the OR of illiteracy for dementia to be 3.67, then the PAF of illiteracy for incident dementia could decrease much further, from 60.02% to 54.41% in 2020 and from 48.49% to 32.00% in 2050, which would save the national economy 94.27 billion USD by 2050.

Lastly, in Africa the OR of illiteracy for dementia has been reported to range from 1.22 to 4.89. Assuming the OR of illiteracy for dementia to be 1.22, a 50% reduction of illiteracy in African people under 65 years of age would predict a reduction in the PAF of illiteracy for incident dementia from 12.29% to 10.06% in 2020 and from 6.06% to 3.12% in 2050. This would be expected to save the national economy 17.46 billion USD in dementia care costs by 2050. If we assumed the OR of illiteracy for dementia to be 4.89, then the PAF of illiteracy for incident dementia could decrease much further from 71.24% to 66.42% in 2020 and from 53.27% to 36.30% in 2050, which would save the national economy 77.75 billion USD by 2050.

DISCUSSION

In this study, we estimated the PAF of illiteracy for incident dementia in a number of regions that span multiple continents. We also estimated the potential cost savings to the national economy in dementia management after reducing illiteracy to a more manageable extent. Our results have demonstrated that around 16% of dementia cases in South Korea can be attributed to illiteracy in 2015, and that this proportion is expected to decline to around 2% by 2050. This translates to a potential saving of 52 billion USD in dementia care costs by 2050, providing as of 2015 we are able to, theoretically, eradicate illiteracy in Korean people under 65 years of age. Applying this method, a 50% reduction of illiteracy could further reduce the dementia care costs of Latin America, South Asia/The Middle East, and Africa by as much as 71 to 244, 13 to 94, and 17 to 78 billion USD respectively, by 2050.

Several studies have estimated the PAF of Alzheimer’s disease for modifiable risk factors that have been known to show robust evidence of association other than illiteracy. Most notably, Norton et al. [33] have recently reviewed the relevant data and concluded that as of 2010, approximately 30% of the worldwide incidence rate of Alzheimer’s disease can be attributed to the effects of a combination of seven risk factors, including diabetes mellitus, midlife hypertension, midlife obesity, physical inactivity, depression, smoking, and low educational attainment. In another analysis conducted using cohorts from the Rotterdam Study [34], the combined PAF of dementia for overweight and obesity, hypertension, diabetes mellitus, unfavorable cholesterol levels, smoking, and low educational attainment, was reported to be around 20–30%. Although it is difficult to draw direct comparisons between these results and our present study, it could be argued, given individual PAFs of approximately 3–20% for each abovementioned risk factor, that these modifiers are not as influential as illiteracy, which in 2015, had a PAF for dementia of approximately 5–70%. That is to say, the reduction of illiteracy, among other preventative measures, is probably the most powerful way of reducing the prevalence of dementia in developing countries.

Additionally, among numerous impacts after reducing illiteracy, the dementia-preventive effect is noteworthy, particularly in Latin America. The World Literacy Foundation has estimated the global economic cost of illiteracy using a mathematical formula from the United Nations Educational, Scientific and Cultural Organization (UNESCO) [35]. According to this calculation, illiteracy has cost the South Korean, Latin American, South Asian/Middle Eastern, and African governments approximately 31, 68, 238, and 4,552 billion USD, respectively in 2010. Since illiteracy rates decrease over time and savings in annual dementia care costs begin to build steadily, after reducing illiteracy as of 2015, we were able to predict a role of the “dementia-preventative effect” of reducing illiteracy that will have much more significancethan any other socioeconomic benefit. This is especially the case in Latin America where a 50% reduction in illiteracy as of 2015 would result in a socioeconomic cost of illiteracy of less than 30 billion USD by the year 2050, almost half of which is the potential amount saved annually by preventing dementia.

We could also argue that, by delaying the onset of dementia through the reduction of illiteracy, the mortality rate of people with dementia would probably not be affected [36]. This fact can be explained by the CR hypothesis which posits that the educational attainment does not affect the neurodegenerative process per se, rather it may affect the clinical manifestation of the disease. This argument was further supported by Helmer et al. [37] who claimed that the degree of education was not significantly associated with the mortality rate of people with dementia. However, there also exits controversy concerning this relationship: some have found high level of education is associated with the high mortality rate [38, 39], while others have failed to replicate this finding or even proposed the opposite [40, 41]. Again, according to the CR hypothesis, degenerative cerebral lesions of highly educated people might be far more advanced than those of less educated people at the time of diagnosis of dementia. This seems to explain why high level of education was associated with increased mortality, though it is not without a possible detection bias that people with low educational level with low premorbid cognitive function could be diagnosed with dementia earlier than highly educated people. That leads us to say that reducing the illiteracy rate may not necessarily result in increased survival of patients with dementia who would have required more healthcare resources on average.

From this data, we believe that government agencies can be provided with tangible outcomes with specific estimates of possible savings accompanying the reduction of illiteracy, thus endowing initiatives to seriously consider the importance of illiteracy in the management of dementia. As a practical matter, in view of the difficulties accompanied with teaching in developing countries, we need to reform primary school systems in these areas to be able to effectively deliver the monetary aid since distorted educational budgets bring about wasteful allocation of funds [42]. Paying sufficient incentives to teachers in order to reduce absenteeism, developing typical student-matched curriculums, or providing computer-based educational resources are good examples of inexpensive measures that can be implemented rapidly.

This sort of literacy campaigns could possibly play a key role in the management of the dementia because prevention-oriented models are becoming the focus of attention in the face of repeated failures of the curative drug development. More importantly, there is a growing recognition that the mental stimulation may increase cognitive and brain reserves not only in adult or late life, but also throughout childhood and youth [43]. Namely, the reserves are deemed as the summation of inputs of lifetime experiences, thus making the concept of a ‘life-course approach’ to the etiology of late-onset dementia promising [44]. To date, the definite time windows when the education, particularly about literacy, has its greatest protective effect on the risk of dementia is still unknown. However, recent studies suggested that high early-life school performance is associated a reduced risk of dementia, which is independent of late-life educational attainment [45]. Even more, it is also protective of development of dementia, especially when sustained through adulthood complex work environment, and this effect persists even in the absence of late-life educational stimulation [46]. That said, these literacy campaigns might most benefit from focusing on the childhood, adolescent, or middle-aged population. However, thus far, most dementia-preventive strategies of advocacy organizations are primarily concerned with risk factors of old age people [47, 48]. Though, in their recent reports, Alzheimer’s Disease International (ADI) and World Health Organization (WHO) [49, 50] stressed the ‘low education in early life’ as one of the strongest causal association with dementia, these groups did not reached to the point where a specific action plan to educate young people can be set up. This may partly due to the problems related to the lack of public awareness, limited interest of policy makers, and allocating scarce research funds. We believe that if these groups are provided with forthcoming studies concerning illiteracy, they would do well to advocate for the health of young brains. In this article, we theoretically reduced the illiteracy rate of people aged 65 years or younger to demonstrate the effect of targeting these young population. Further studies will be necessary to define the most efficient target population for aiming the literacy campaign to reduce the prevalence of the dementia.

There is considerable ambiguity that needs to be emphasized in this study. First, the “trend scenario” [20] model we adopted to project literacy rates onto the future may not be appropriate for the best guess scenario. By assuming constant transition to literacy without reaching 100%, this model allows simplified translation of cross-sectional data into longitudinal data, but at the same time, cannot take into account country-specific factors, such as the effects of migrating populations, government stated plans regarding literacy, or other political situations that might disrupt the education of the general population. Nevertheless, our model produced adult (aged 15 years and older) literacy rates of 2015, which were similar to those reported by the UNESCO (89.3% versus 92.8% in Latin America, 64.8% versus 70.4% in South Asia/The Middle East, and 60.1% versus 63.6% in Africa [51]). Second, we cannot be confident that we have ascertained precisely how long it will take, after acquiring literacy, to achieve cognitive reserves, which are sufficient to exert a protective effect against the future development of dementia. We presumed that after 5 years of literacy experience, there will be satisfactory amount of reserves based on the previous study about the cerebral structural changes of late-literates [22]. However, we do not have any data about these changes before or after 5 years according to the cognitive abilities of participants, thus making this assumption unfounded. Nevertheless, we believe that this assumption is the closest possible one with insufficient evidence at the present time. Third, the CR hypothesis itself has its own drawback. If we presume that the neurodegenerative process begins at a fixed time point in one’s life, then we would expect that higher CR is associated with a delayed onset and longer survival. However, if we presume that the neurodegenerative process begins at a variable time, and the process is insidious and gradual, then the CR would have no substantial effect on the age of onset of clinical manifestation of patients with dementia [43]. Moreover, some researchers also argue that, instead of the “brain reserve hypothesis,” a “brain battering” model may instead explain the increased risk of dementia for some of illiterate or poorly educated populations [43]. This alternative explanation arose from the autopsy findings of patients with dementia that less educated patients developed dementia later and survived longer, while the rate of cognitive decline and burden of neuropathology were not differ among all educational groups [52]. Interestingly, the less educated group showed more cerebrovascular lesions. Thus this model concluded that greater educational attainment does not affect the course of dementia, while lower educational status is associated with the cerebral infarcts. It assumes that individuals with lower educational attainment and lower socioeconomic status would be exposed to more toxins, have poorer access to quality health care and keep unhealthy lifestyle, all of which would make them vulnerable to brain lesions leading to developing dementia. In other words, it is probable that there is a specific point in life after which there would be no cognitively beneficial effects of teaching languages. Fourth, since all of the studies included here have used self-reporting to decide whether the subject is literate or not, there can be substantial room for inaccuracies in this respect. Furthermore, this work has an important shortcoming in that we selected certain ORs associated with illiteracy of one country, including unadjusted ones, to represent the OR of the whole continent that it belongs to. Since many developing countries within these continents are not capable of conducting reliable nationwide prevalence studies, we only managed to acquire a handful of the ORs adjusted for several comorbidities which are shown in Table 1. To get a conceivable range of PAFs and savings of dementia care costs, we included the ORs which are not adjusted and presented the outcomes under both the highest and lowest ORs. Without sufficient data, we would argue that this is a reasonable way of estimating the PAF of illiteracy for dementia. Further research into the association between illiteracy and the risk of dementia adjusted for various comorbidities will provide much more realistic outcomes in this regard.

In conclusion, even public policies involving relatively inexpensive, and short-term education aimed solely at reducing illiteracy in childhood, adolescent, or middle-aged people, could potentially play an important role in the primary prevention of dementia. Governments, therefore, will need to address this issue in a purposeful and systematic manner in the future.