Cognitive impairment has been associated with low vitamin B12(Reference Green, Allen and Bjørke-Monsen1) and low folate status(Reference Ramos, Allen and Mungas2). In addition, some previous observational data have suggested that older adults with low vitamin B12 and high folate status had higher risks of anaemia and cognitive impairment or decline(Reference Morris, Jacques and Rosenberg3–Reference Moore, Ames and Mander5); however, the causal relevance of these associations remains uncertain.
Vitamin B12 and folate participate in one carbon metabolism, critical for the formation of DNA and RNA, neurotransmitters, membrane lipids and proteins. Because of these functions, there has been growing interest in the possibility that age-related conditions, in particular loss of cognitive function, may partly be explained by inadequate status of these vitamins(Reference Raman, Tatsioni and Chung6–Reference Vogel, Dali-Youcef and Kaltenbach9). Vitamin B12 deficiency (plasma vitamin B12 < 148 pmol/l) is associated with severe neurological consequences, including sensory and motor neuropathy, altered mental status and cognitive decline(Reference Green, Allen and Bjørke-Monsen1). Vitamin B12 deficiency also reduces the availability of folate for DNA synthesis(Reference Green, Allen and Bjørke-Monsen1).
Since the introduction of mandatory fortification of staple foods with folic acid to reduce neural tube defects (NTD), high folate status (>45·3 nmol/l) is frequently observed in countries such as the USA and Canada. Recent research has raised concerns regarding the adverse implications of simultaneous low vitamin B12 and high folate status given that high-dose folic acid treatment was shown to temporarily mask clinical symptoms in persons with pernicious anaemia(Reference Ross, Belding and Paegel10).
While some studies have reported that older adults with low vitamin B12 and high folate status had higher risks of anaemia, cognitive impairment and decline(Reference Morris, Jacques and Rosenberg3–Reference Moore, Ames and Mander5), other large studies conducted in diverse populations have not replicated such adverse associations(Reference Clarke, Sherliker and Hin11–Reference Doets, Ueland and Tell13). There is considerable heterogeneity among the published studies, including differences in vitamin status cut-offs, biochemical methodologies and the instruments used to assess cognitive function(14). Likewise, large-scale randomised controlled trials have failed to demonstrate any beneficial effects of supplementation with vitamin B12 or folic acid on global or domain-specific cognitive function(Reference Eussen, de Groot and Joosten15).
Ireland has one of the highest rates of neural tube defects in Europe(Reference Botto, Lisi and Robert-Gnansia16), and supplementation initiatives have been shown to be ineffective(Reference McDonnell, Delany and O’Mahony17), yet one major obstacle impeding population-based food fortification is the concern that high folate status in the presence of low vitamin B12 status could have a detrimental effect on cognitive function in older people. In our population study, The Irish Longitudinal Study on Ageing (TILDA), the prevalence of older people with deficient or low vitamin B12 status (<185 pmol/l) was 12 %(Reference Laird, O’Halloran and Carey18), while high folate status (>45·3 nmol/l) was observed in 9 % of the population(Reference Laird, O’Halloran and Carey18). TILDA affords a unique opportunity to investigate whether there is an association between cognitive performance and variations of vitamin B12 and folate in a large, nationally representative population sample of older adults. We hypothesise that a low vitamin B12 status coupled with high blood concentrations of folate does not confer greater risk to global cognitive performance in this older population group.
Methods
Study population
Cross-sectional data from Wave 1 of TILDA were analysed. TILDA is a nationally representative prospective cohort study on ageing of community-dwelling adults aged 50 years and older, living in the Republic of Ireland. Details of the study design have been reported previously(Reference Cronin, O’Regan and Finucane19–Reference Whelan and Savva21). Briefly, Wave 1 took place between October 2009 and February 2011. Eight thousand hundred and seventy-four adults aged ≥50 years from 6279 households completed a computer-aided personal interview, representing a response rate of 62 %. Of the total sample, 84·6 % completed the self-completion questionnaire (n 7196) and 72·1 % (n 5895) participated in a health assessment which included blood measurements. Plasma vitamin B12 measurements were available for 88·5 % (n 5219) and plasma folate for 90·7 % (n 5350) of participants. Missing data were due to insufficient sample and/or assay failures. Individuals who did not complete both standardised assessments of cognitive function, the Mini Mental State Examination (MMSE)(Reference Folstein, Folstein and McHugh22) and Montreal Cognitive Assessment (MoCA)(Reference Nasreddine, Phillips and Bédirian23), were not included in this analysis, giving a final analytical sample of 3871 (Fig. 1). Ethical approval was obtained from the Faculty of Health Sciences Research Ethics Committee of Trinity College Dublin. All participants were fully informed of the study and gave written consent.
Fig. 1. Flow diagram – analytical sample design and selection. TILDA, The Irish Longitudinal Study on Ageing; MMSE, Mini Mental State Examination; MoCA, Montreal Cognitive Assessment; B12, vitamin B12.
Assessment of cognitive function
Global cognitive function was assessed using the MMSE and MoCA. The MMSE is a brief 30-point test that is used to screen for cognitive impairment, commonly used in clinical practice to screen for dementia and assesses orientation, recall, attention, calculation, language abilities and visuospatial ability. The MoCA, also used the screen for dementia, is particularly sensitive to mild cognitive deficits(Reference Nasreddine, Phillips and Bédirian23,Reference Trzepacz, Hochstetler and Wang24) , with additional assessments of executive function, abstraction and visuospatial ability. Both tools have scores ranging from 0 to 30.
Analysis of B-vitamins
A non-fasting blood sample was collected by venepuncture into one 10 ml K2EDTA tube (BD, Becton, Dickinson Limited) for immediate analysis and two 10 ml EDTA (BD, Becton, Dickinson Limited) tubes for long-term storage (Sarstedt). Samples were kept chilled and were centrifuged (3000 rpm for 15 min). Aliquots of plasma were labelled and stored at –80°C within 48 h of blood collection and stored until required for analysis. The TILDA protocol for blood sample collection, processing, and storage is detailed elsewhere(Reference Whelan and Savva21). Plasma vitamin B12 and folate concentrations were determined by microbiological assay; Lactobacillus leichmannii (L. leichmanii (delbrueckii) (NCIB 12519, ATCC 43787)), limit of detection 3·7 pmol/l(Reference Kelleher and Broin25) and Lactobacillus casei (L. rhamnosis (NCIB 10463; ATCC 27773)), limit of detection 0·6 nmol/l(Reference Molloy and Scott26), respectively. Inter-assay CV for both methods were <10·9 %.
Other covariates
Sociodemographic characteristics examined included age, sex, educational attainment, mean asset wealth and habitation status. Lifestyle characteristics comprised self-reported physical activity (International Physical Activity Questionnaire), smoking status, alcohol consumption, depressive symptoms (Center for Epidemiologic Studies Depression Scale)(Reference Radloff27) and history of stroke. Use of medications (up to twenty prescription and non-prescription medications, including food supplements) was classified by World Health Organization, Anatomical Therapeutic Chemical classification codes.
Anthropometric measurements included height, weight, BMI and handgrip strength (average two recordings from dominant hand using a dynamometer). Hypertension was defined as systolic blood pressure ≥ 140 mmHg or diastolic blood pressure ≥ 90 mmHg and/or antihypertensive medication use. Diabetes was defined as diabetes or prediabetes by measurement of glycated Hb (HbA1c)(28), self-reported diabetes diagnosis and/or use of anti-glycaemic medications.
Exclusion criteria
Individuals without a health assessment and blood sample were excluded (n 3248). Those for whom vitamin B12 and/or folate concentrations could not be measured, or whose measurement values exceeded the 99th percentiles and participants in receipt of vitamin B12 injections, were omitted from the analysis (n 450). In order to rule out participants with possible mild cognitive impairment, additional exclusionary criteria were applied to both MMSE and MoCA results. Those with MMSE scores of <24(Reference Lezak, Howieson and Loring29) and MoCA below the age and education normative cut-offs of Rossetti et al. (Reference Rossetti, Lacritz and Cullum30) were also omitted (n 605).
Statistical methods
While there is no international consensus regarding the cut-offs for both vitamin B12 and folate status, we adopted the widely used cut-offs identified in the literature. For plasma vitamin B12 (pmol/l), these included <148 (deficient); 148–<185 (low status); 185–<258 (low/normal status); >258–>601 (normal) and >601 (high)(Reference Bailey, Carmel and Green31,Reference Arendt and Nexo32) . For plasma folate (nmol/l), the cut-offs included 10–23·0 (low/normal) and >23·0–45·3 (normal) >45·3 (high)(Reference Miller, Garrod and Allen12,Reference Selhub, Morris and Jacques33) . To compare with previous studies, a four-level categorical variable was created on the basis of vitamin B12 (< or ≥258 pmol/l)(Reference Bailey, Carmel and Green31,Reference Arendt and Nexo32) and folate concentrations (≤ or >45·3 nmol/l)(Reference Pfeiffer, Caudill and Gunter34) and was defined as: (i) normal vitamin B12/normal folate (≥258 pmol/l, ≤45·3 nmol/l), (ii) normal vitamin B12/high folate (≥258 pmol/l, >45·3 nmol/l), (iii) low vitamin B12/high folate (<258 pmol/l, >45·3 nmol/l) and (iv) low vitamin B12/normal folate (<258 pmol/l, ≤45·3 nmol/l).
Non-normally distributed continuous variables were described as medians and interquartile ranges for each group, stratified by sample characteristics, biomarkers and cognitive tests. These were compared using Kruskal–Wallis and Mann–Whitney U tests for pairwise comparisons. Categorical variables were compared using χ 2 tests or Fisher’s exact test, as appropriate. A false discovery rate correction for multiple comparisons was applied (P < 0·013, achieving false discovery rate q < 0·05) to adjust for the number of individual tests carried out (i.e. q indicates that the average alpha level across the family of tests does not exceed 0·05, following correction).
Associations of vitamin B12 and folate with cognitive function and other covariates were explored in a series of regression analyses. The relationship between cognitive function and groups of vitamin B12 and folate status was analysed with models appropriate for the distribution of the outcome variables, MMSE (continuous) and MoCA (continuous). Visualisation of the data and detailed inspection of model residuals demonstrated that neither outcome was normally distributed after appropriate removal of outliers. Therefore, MMSE and MoCA were analysed using negative binomial regression models (due to overdispersion) with the outcome representing the count of errors made during the test (30 minus the raw score) due to their left-skewed distributions, consistent with previous reports using TILDA data(Reference Feeney, O’Halloran and Kenny35).
The results of these analyses are presented as the incidence rate ratio for each model, relative to the reference category, normal vitamin B12/normal folate, as previously performed in this cohort(Reference Donoghue, Feeney and O’Leary36). For both cognitive tests, three models were investigated: unadjusted (model 0); adjusted for sociodemographic covariates – age, sex, educational attainment, mean asset wealth and habitation status (model 1) and adjusted for sociodemographic, lifestyle, health indicators and medications – physical activity, smoking, alcohol consumption, BMI, mean grip strength, hypertension, diabetes, depressive symptoms, folic acid supplement use, proton pump inhibitor, statin and thyroid medication use and history of previous stroke (model 2).
These analyses were then repeated in the sample aged ≥60 years (n 2070), for more direct comparison with other studies in this field. Further, a sensitivity analysis was performed including the participants with lower cognitive scores, aged ≥50 and ≥60 years, omitted from the main analysis.
Additionally, quintiles of vitamin B12 concentrations were created based on plasma vitamin B12 concentrations in those aged ≥60 years. An interaction term was included to examine whether the association between plasma folate (as a continuous variable) differed between quintiles of vitamin B12, using the highest quintile of vitamin B12/folate (Q5) as the reference category (the reverse instantiation was also analysed, that is, continuous vitamin B12 and quintiles of folate).
Population weights were calculated for the regression analyses by comparing the TILDA population with the 2011 Irish Census with respect to age, sex, educational attainment, marital status and urban v. rural residence (weights reflected the reciprocal of the selection probability). Prevalence estimates, descriptive statistics and model coefficients were adjusted using modified base weights that accounted for survey non-response, non-attendance at the health assessment component of the study and whether a blood sample was provided. The staged, stratified and clustered sample design of TILDA was considered when estimating confidence intervals and standard errors. Analyses were carried out using STATA 15 (StataCorp).
Results
Baseline characteristics of the study sample
The study design and exclusion criteria are detailed in Fig. 1. The mean age was 61·8 years (sd 8·8) (range 50–98 years); 50·7 % were female (n 3871). Median (interquartile ranges) MMSE and MoCA scores were 29 (28–30) and 26 (24–28), respectively. Vitamin B12 concentrations ranged from 45·4 to 719 pmol/l, with 4·8 % (95 % CI 4·1, 5·7) classified as deficient (<148 pmol/l) and 25·5 % (95 % CI 24·1, 27·0) with low vitamin B12 status (148–258 pmol/l). Folate concentrations were 0·9–82 nmol/l, 5·7 % (95 % CI 4·9, 6·7) had high folate status (45·3–60 nmol/l), with an additional 3·3 % (95 % CI 2·8, 4·0) classified as very high folate status (>60 nmol/l), similar to rates observed in our full population(Reference Laird, O’Halloran and Carey18). The prevalence of normal vitamin B12/normal folate status was 62 % (95 % CI 60·2, 63·8), normal vitamin B12/high folate was 7·6 % (95 % CI 6·6, 8·8), low vitamin B12/high folate was 1·5 % (95 % CI 1·1, 2·0) and low vitamin B12/normal folate was 28·9 % (95 % CI 27·3, 30·6).
Sample characteristics, by vitamin B12 and folate groups
Sample characteristics, stratified by vitamin B12 and folate groups, are detailed in Table 1, with additional sample characteristics displayed in online Supplementary Table S1. All groups were older than the reference category, normal vitamin B12/normal folate, and were more likely to be hypertensive. When compared with the reference category, subjects with normal vitamin B12/high folate were more likely to be female, have above average household wealth, lower mean grip strength and BMI and were less likely to be categorised as obese. This group was also more likely to report taking a folic acid supplement and had better global cognitive scores than the reference group. Those with low vitamin B12/high folate status were more likely to be obese, have above average household wealth and be hypertensive and diabetic relative to the reference category. Subjects with low vitamin B12/normal folate status had lower levels of educational attainment, higher overall BMI and were more likely to smoke and to be obese, hypertensive, diabetic and thus to use metformin (online Supplementary Table S1). These groups were further examined by age group (online Supplementary Table S2), though no additional relevant significant findings were observed.
Table 1. Basic sample characteristics, grouped by concentrations of vitamin B12 (B12) and folate†
(Numbers and percentages; medians and interquartile ranges (IQR))
MMSE, Mini Mental State Examination; MoCA, Montreal Cognitive Assessment.
* Statistically significant after adjustment for multiple comparisons (P < 0·013).
† Non-normally distributed continuous variables are described as medians and IQR (all such values) and compared using the Kruskal–Wallis test as appropriate, and the Mann–Whitney U test for pairwise comparisons with the reference category. Categorical variables were compared using χ 2 tests of Fisher’s exact test, as appropriate.
Associations with combinations of vitamin B12 and folate status
Fig. 2 and online Supplementary Table S3 display the associations between the groups of vitamin B12 and folate status in those aged ≥50 and ≥60 years for MMSE. Relative to normal vitamin B12/normal folate, having low vitamin B12/high folate was not associated with a significant increase in MMSE error rates for models 0, 1 or 2. Further, for all three models, there were statistically significant decreases in MMSE error rates for those in the normal vitamin B12/high folate group in both the ≥50 and ≥60 years groups (model 2:0·84 (95 % CI 0·71, 1·00); P = 0·047 and 0·82 (95 % CI 0·68, 0·99); P = 0·043, respectively). Fig. 3 and online Supplementary Table S3 show the associations for MoCA. Similarly, low vitamin B12/high folate status was not associated with increases in error rates in the three models tested. However, having normal vitamin B12/high folate was associated with significant decreases in MOCA error rates for models 0, 1 and 2, respectively, in those aged ≥50 years (0·91 (95 % CI 0·83, 0·99); P = 0·031, 0·92 (95 % CI 0·85, 1·00); P = 0·040 and 0·91 (95 % CI 0·85, 1·00); P = 0·042) and ≥60 years (0·87 (95 % CI 0·78, 0·97); P = 0·014, 0·91 (95 % CI 0·83, 1·00); P = 0·044 and 0·89 (95 % CI 0·80, 1·00); P = 0·025). Results of the full multivariate analyses for models 1 and 2 are presented in online Supplementary Tables S4 and S5. Sensitivity analysis indicated that those in the low vitamin B12/high folate groups for each model and in the subgroup aged ≥60 years were not more likely to make errors with the MMSE or MoCA compared with those within the reference category (online Supplementary Table S6).
Fig. 2. Associations of vitamin B12 (B12) and folate status and global cognition (Mini Mental State Examination (MMSE)) – aged ≥ 50 and ≥ 60 years (incidence rate ratios (IRR) and 95 % confidence intervals). (
), Aged ≥50 years, model 1 (Sociodemographic) (MMSE error rate and vitamin B12 and folate groups with controls for age, sex, educational attainment, mean asset wealth and habitation status). (
), Aged ≥50 years, model 2 (Sociodemographic, Lifestyle, Health Indicators and Medications) (model 1 + controls for physical activity, smoking status and alcohol consumption, BMI, grip strength, hypertension status, diabetes status and depressive symptoms, folic acid supplement use, proton pump inhibitor, statin and thyroid medication use and history of stroke). (
), Aged ≥60 years, model 1 (Sociodemographic) (MMSE error rate and B12 and folate groups controls for age, sex, educational attainment, mean asset wealth and habitation status). (
), Aged ≥60 years, model 2 (Sociodemographic, Lifestyle, Health Indicators and Medications) (model 1 + controls for physical activity, smoking status and alcohol consumption, BMI, grip strength, hypertension status, diabetes status and depressive symptoms, folic acid supplement use, proton pump inhibitor, statin and thyroid medication use and history of stroke).
Fig. 3. Associations of vitamin B12 (B12) and folate status and global cognition (Montreal Cognitive Assessment (MoCA)) – aged ≥50 and ≥60 years (incidence rate ratios (IRR) and 95 % confidence intervals). (), Aged ≥50 years, model 1 (Sociodemographic) (MoCA error rate and vitamin B12 and folate groups with controls for age, sex, educational attainment, mean asset wealth and habitation status). (), Aged ≥50 years, model 2 (Sociodemographic, Lifestyle, Health Indicators and Medications) (model 1 + controls for physical activity, smoking status and alcohol consumption, BMI, grip strength, hypertension status, diabetes status and depressive symptoms, folic acid supplement use, proton pump inhibitor, statin and thyroid medication use and history of stroke). (), Aged ≥60 years, model 1 (Sociodemographic) (MoCA error rate and vitamin B12 and folate groups with controls for age, sex, educational attainment, mean asset wealth and habitation status). (), Aged ≥60 years, model 2 (Sociodemographic, Lifestyle, Health Indicators and Medications) (model 1 + controls for physical activity, smoking status and alcohol consumption, BMI, grip strength, hypertension status, diabetes status and depressive symptoms, folic acid supplement use, proton pump inhibitor, statin and thyroid medication use and history of stroke).
Associations with quintiles of vitamin B12 and continuous folate concentrations
Additional models, in subjects aged ≥60 years, examined folate concentrations on a continuum, interacted with quintiles (Q1–Q5) of vitamin B12 (Q1 vitamin B12, 45·4–225 pmol/l (n 409); Q2 225–289 pmol/l (n 427); Q3 289–350 pmol/l (n 408); Q4 350–428 pmol/l (n 415); Q5 428–719 pmol/l (n 412)). These demonstrated there was no evidence of interactions with respect to MMSE and MoCA error rates (Fig. 4). For MoCA, model 2 revealed a small yet statistically significant decrease in the rate of errors for those with the highest vitamin B12 concentrations (online Supplementary Table S7). The reverse instantiation was also tested (i.e. vitamin B12 as the continuous variable, interacted with quintiles of folate). This analysis showed no association of error rates with decreasing concentrations of vitamin B12 (online Supplementary Table S8).
Fig. 4. Associations between concentrations of folate (continuous (Cont)) and quintiles (Q) of vitamin B12 and global cognition – aged ≥60 years (incidence rate ratios (IRR) and 95 % confidence intervals). (), Aged ≥60 years, model 1 (Sociodemographic) (Global cognition measure error and vitamin B12 and folate concentrations with controls for age, sex, educational attainment, mean asset wealth and habitation status). (), Aged ≥60 years, model 2 (Sociodemographic, Lifestyle, Health Indicators and Medications) (model 1 + controls for physical activity, smoking status and alcohol consumption, BMI, grip strength, hypertension status, diabetes status and depressive symptoms, folic acid supplement use, proton pump inhibitor, statin and thyroid medication use and history of stroke). MMSE, Mini Mental State Examination; MoCA, Montreal Cognitive Assessment.
Discussion
The present study estimated the prevalence of groups of vitamin B12 and folate status in a nationally representative sample of older adults in Ireland and investigated the hypothesis that low vitamin B12 status in conjunction with high folate status is associated with increased risk of cognitive decline in older people(Reference Morris, Jacques and Rosenberg3–Reference Moore, Ames and Mander5). We demonstrated that the 1·5 % of the population with low vitamin B12/high folate status did not have any adverse associations with cognitive performance compared with those with normal status for both B-vitamins. In contrast, the present study demonstrated that higher concentrations of folate were associated with small, but statistically significant higher scores for MMSE and MoCA performance. In contrast to previous research using single measures of global cognition in older cohorts, we employed two clinically validated methods of global cognitive function measurement simultaneously.
There are a number of possible explanations as to why our findings diverge with those of the National Health and Nutrition Examination Survey (NHANES), the Framingham Heart Study and three combined cohorts of Moore and colleagues (2014)(Reference Morris, Jacques and Rosenberg3–Reference Moore, Ames and Mander5). In NHANES (n 1457), concentrations of high plasma folate (>59 nmol/l) with low vitamin B12 (plasma B12 < 148 pmol/l or methylmalonic acid > 0·21 mmol/l) (n 42) were associated with poorer cognitive performance cross-sectionally. The latter study was conducted in the USA, a region implementing mandatory folic acid fortification and where the use of supplements containing folic acid was common, leading to extremely high concentrations of plasma folate and the presence of circulating unmetabolised folic acid(Reference Morris, Jacques and Rosenberg3,Reference Pfeiffer, Sternberg and Fazili37,Reference Qi, Crider and Tinker38) . Thus, very high concentrations (plasma folate >59 nmol/l) were observed in 20·7 % of the NHANES cohort, compared with just 3·3 % (plasma folate ≥ 60 nmol/l) in the present study. The use of folic acid supplements was uncommon amongst older adults in Ireland, ranging between 7·2 and 13·7 % amongst the two groups with the highest status. In addition, the Digit Symbol-Substitution subtest of the Wechsler Adult Intelligence Scale III was used in NHANES to examine cognitive performance in attention and processing speed, while the present study assessed global cognitive performance using both MMSE and MoCA. From the Framingham Heart Study(Reference Morris, Selhub and Jacques4), the data of which pertained to pre-mandatory fortification in the USA (1986–1990), Morris et al. (Reference Morris, Selhub and Jacques4) reported faster rates of decline in MMSE were associated with vitamin B12 concentrations of <257 pmol/l and plasma folate of >20·2 nmol/l – after 8-year follow-up (n 24). While the vitamin B12 cut-off used was similar to the present study, we applied a higher threshold to define high folate status and used a cognitively intact sample. Furthermore, there were twice as many participants included in the present analysis.
Moore et al.(Reference Moore, Ames and Mander5) also suggested that high folate or folic acid supplements may be detrimental to cognition in older people with low vitamin B12 concentrations (n 1354)(Reference Moore, Ames and Mander5). Compared with the present study, a similar vitamin B12 cut-off (<250 pmol/l) was used, and red cell folate of >1594 nmol/l, which approximate to those employed in the Framingham Heart Study investigation and had thirty-nine participants in the low vitamin B12/high folate group. However, the sample was recruited from three separate clinical cohorts in Australia, a region also implementing mandatory folic acid fortification. In addition, 480 subjects had established Alzheimer’s disease and 187 had confirmed mild cognitive impairment. This is in contrast with our study, all of whom were free living and cognitively robust (n 3871). In addition, Moore et al.(Reference Moore, Ames and Mander5) used the binary outcome of MMSE of <24 to define cognitive impairment, whereas our study used MMSE and MoCA as a continuous outcome to investigate the range of risk to cognitive performance.
The findings of the present study are consistent with other studies using European population data; those of Clarke et al.(Reference Clarke, Sherliker and Hin11) and Doets et al.(Reference Doets, Ueland and Tell13) showing no association of low vitamin B12/high folate status with an increased risk to cognitive function. In addition, the Sacramento Area Latino Study on Aging in America(Reference Miller, Garrod and Allen12) did not demonstrate a negative association with cognitive function (n 1535) using a vitamin B12 deficiency cut-off of <148 pmol/l and high folate of >45·3 nmol/l, although biochemical anomalies were observed.
To our knowledge, this is the largest such study to use a nationally representative sample of community-dwelling older adults that were cognitively unimpaired, employing rigorous exclusion criteria for cognitive performance to ensure accurate investigation of early cognitive changes and to mitigate potential confounding. Another strength of the present study is the use of two tests of global cognitive performance in parallel. Most other studies in this field have investigated the MMSE or variations of it to determine cognitive performance or impairment. We also measured global cognitive performance using the MoCA, an arguably better screening tool for subtle cognitive decline and mild cognitive deficits(Reference Nasreddine, Phillips and Bédirian23,Reference Trzepacz, Hochstetler and Wang24) . In addition, we are the first study to explore interactions of vitamin B12 and folate concentrations with respect to MoCA and we controlled for a large extent of covariates to account for factors that may affect B-vitamin status, particularly medications and food supplement use.
Consistent with previous studies, the present study was limited by the low numbers of participants in the ‘at-risk’ group. This determined the cut-off for vitamin B12 (<258 pmol/l) utilised in the present study, as there were too few individuals with vitamin B12 concentrations in the deficient range. The present study had only fifty-one participants in the low vitamin B12/high folate group, albeit this number exceeded previous studies addressing this hypothesis. The low number in the group of interest may reflect the large proportion of the older population of Ireland captured by ‘sub-optimal’ but not deficient status, and the inherent selection bias in analysis of the most robust participants in a study population. However, the vitamin B12 cut-off used in the present study is comparable with those used by the Framingham and Moore et al. studies(Reference Morris, Selhub and Jacques4,Reference Moore, Ames and Mander5) . Further, the small number of such individuals included in the present study reflected the application of strict exclusion criteria, and the authors are satisfied that this is representative of the proportion of those aged ≥50 years in this interest group. Similarly, our investigation using continuous concentrations of plasma folate was conducted to compensate for the low proportions of participants with vitamin B12 concentrations below 148 pmol/l.
Another limitation was the cross-sectional nature of the present study, and therefore we cannot predict decline in the at-risk group. Moreover, the present study was limited to the use of plasma vitamin B12 and folate as indicators of status, acknowledging the advantages and disadvantages associated with these biochemical measures (i.e. total vitamin B12 in blood has relatively low sensitivity for estimating vitamin B12 deficiency(Reference Valente, Scott and Ueland39) when compared with other biomarkers, whereas plasma folate can be subject to variations from recent dietary intakes). In addition, dietary intake data at Wave 1 were not ascertained, therefore we could not estimate dietary exposure of both vitamin B12 and folate. It is also important to note that because of the recruitment sampling criteria and rigorous exclusion criteria in the present study, these findings may not be generalisable to non-Caucasians, older adults in institutional care or those with cognitive impairment. Finally, we acknowledge the potential reverse causality and bidirectional relationship between global cognitive performance and vitamin B12 and folate status and that the possible association between B-vitamin status and cognitive impairment in older adults might reflect effects of such vitamins on Alzheimer’s disease or cerebrovascular disease, which are the primary pathogenic processes underlying cognitive impairment in older adults.
In summary, we undertook the present study to assess the possible hazards for cognitive function in older people associated with voluntary folic acid fortification in Ireland and found no evidence that high blood folate concentrations affected the associations of low vitamin B12 status with cognitive impairment. Our inability to detect differences in cognitive performance using these instruments implies that elevated folate status should not cause an increased risk to the present study population. In contrast, our data suggest that higher folate status may be protective of global cognitive function irrespective of vitamin B12 status, in a representative sample of community-dwelling population of older people. This is particularly relevant, as our group has previously determined that in addition to 12 % of older adults in the Republic of Ireland having low or deficient vitamin B12 status, up to 15 % also had deficient or low folate status(Reference Laird, O’Halloran and Carey18) – demonstrating that the policy of voluntary fortification with folic acid is ineffective for older people in the community.
These findings have significant implications for public policy on folic acid fortification in Ireland and elsewhere in Europe and provide reassurance to both healthcare providers and policy makers that high folate status is not associated with increased risk of cognitive impairment in older people.
Acknowledgements
We are grateful to all the TILDA participants for their engagement with the project.
This work was supported by the Irish Department of Agriculture, Food and the Marine through the grant 13F492: The Nutritional Biomarker Database Enhancement Initiative of (‘BIO-TILDA’ 2013–2015). Original funding for TILDA was provided by The Atlantic Philanthropies, the Irish Government and Irish Life plc. The funders played no role in designing or conducting the study or in the collection, management, analysis or interpretation of the data, nor did they have any input into the preparation, review or approval of the manuscript.
The author responsibilities were as follows: D. M. A. O., E. J. L., R. A. K. and A. M. M. designed the research; D. M. A. O., E. J. L., D. C. and A. M. O. collected and analysed the data; D. M. A. O., E. J. L. and D. C. performed statistical analysis; D. M. A. O., E. J. L., D. C., R. A. K. and A. M. M. wrote the initial draft of the manuscript; D. M. A. O., E. J. L., D. C., A. M. O., R. C., R. A. K. and A. M. M. critically revised the manuscript; and all authors had responsibility for accuracy of the final content and read and approved the final manuscript.
The authors declare that there are no conflicts of interest.
Supplementary material
For supplementary material referred to in this article, please visit https://doi.org/10.1017/S0007114520001427
