Introduction

Cognitive decline occurs across the lifespan, with the domains of memory and attention particularly vulnerable to cognitive deterioration. Population studies have revealed that vitamin deficiency (La Rue et al. 1997; Quadri et al. 2005), cardiovascular disease (Breteler et al. 1994), elevated homocysteine (Elias et al. 2006) and cellular metabolic dysfunction (Berr et al. 1998) can be considered risk factors for cognitive decline, dementia, and other pathological conditions of ageing. Growing evidence indicates that dietary supplementation with selected vitamins and minerals may be capable of improving these parameters (Block et al. 2007) and as a consequence may also influence cognitive function (Huskisson et al. 2007).

Adequate vitamin intake is vital for the optimum functioning of the brain and nervous system. To maintain normal levels of homocysteine, vitamin B6, vitamin B12 and folic acid are required in the methylation of homocysteine to methionine. In turn, methionine is required in the synthesis of sadenosylmethionine (SAM), the sole donor for methylation reactions in the brain. Products of these reactions include the neurotransmitters dopamine, norepinephrine and serotonin, as well as proteins, phospholipids, DNA, and myelin (Selhub et al. 2000). Vitamins A, C, and E, selenium and coenzyme Q10 exert antioxidant effects and protect neural tissue from aggression by free radicals (Huskisson et al. 2007). In the body, these vitamins do not operate in isolation and may exert more potent effects when administered in combination, such as in the form of a multivitamin supplement.

Recent investigations have focused on the potential for multivitamins to enhance cognitive function. Several of these trials in the elderly have revealed only very small or no cognitive improvements (Cockle et al. 2000; Wolters et al. 2005; McNeill et al. 2007). However, it is plausible that these negative findings may have partially been influenced by the choice of cognitive measures employed in these trials. For instance, in a study of healthy seniors, there were no improvements on digit span forward and verbal fluency measures associated with 12 months multivitamin treatment (McNeill et al. 2007). As the primary aim of this study was to investigate infection, only two secondary cognitive measures of short-term memory and executive function were included in the protocol. Similarly, Wolters et al. (2005) did not identify any cognitive benefits of 6 months multivitamin treatment in elderly women. Again, cognitive assessment was restricted to estimates of verbal intelligence and symbol search, instruments which have been suggested to be less sensitive to subtle nutraceutical effects than computerized measures of attention or non-verbal memory (Haskell et al. 2008). By contrast, recent investigations conducted in young to middle-aged adults have identified multivitamin related benefits on a computerized multi-tasking framework (Haskell et al. 2010) and a computerized mental subtraction task (Kennedy et al. 2010), indicating that computerized measures of fluid intelligence may be particularly responsive to the cognitive enhancing effects of multivitamins.

In addition to combined vitamin formulations, there is evidence that selected herbals possess cognitive enhancing properties. In healthy elderly, benefits to declarative episodic memory have been reported after 6 weeks Ginkgo biloba treatment (Mix and Crews 2002) and to verbal memory following 12 weeks treatment (Burns et al. 2006). Other plant extracts containing flavonoid components have also been suggested to improve cognition (Macready et al. 2009). Positive results regarding herbals combined with vitamins has been obtained from a recent 4-month trial in elderly subjects which examined the effects of a complex antioxidant blend consisting of vitamins, minerals, amino acids, lipids and herbal extracts, all with antioxidant properties (Summers et al. 2010). Herbal components of this formula included flavonoids, Ginkgo biloba, grape seed and gotu kola—ingredients which have gained interest as potential modulators of cognitive function (Macready et al. 2009; Spencer 2009). Treatment with the antioxidant blend was associated with improvements on cognitive measures of word recall and a paired-associates word task. The treatment also reduced serum homocysteine levels in a subset of the sample.

Although the cognitive improvements identified by Summers et al. (2010) were promising, without the examination of treatment effects on other biological indices such as blood measures of vitamin status, oxidative stress, inflammation or cardiovascular parameters, the mechanisms of action can only be speculated. Whilst a 3-year trial designed to reduce homocysteine through folate supplementation demonstrated improvements to memory, as assessed by a verbal learning task, information processing speed and sensorimotor speed in seniors aged 50–70 years (Durga et al. 2007), there is little evidence that in healthy elderly, lowering homocysteine, through shorter term B vitamin supplementation alone, improves cognition (Lewerin et al. 2005; Eussen et al. 2006; McMahon et al. 2006; Balk et al. 2007). These findings indicate short-term reductions in homocysteine may not be solely responsible for cognitive improvements in healthy elderly associated with multivitamin intake.

Trials without a cognitive component have revealed that chronic multivitamin supplementation is capable of reducing other risk factors for cognitive decline and cardiovascular disease, including levels of oxidative stress and inflammation. In studies of cardiovascular parameters, multivitamins have been demonstrated to reduce platelet activation (Salonen et al. 1991; Arnaud et al. 2007), concentrations of the C-reactive protein marker of inflammation (Church et al. 2003), and to increase HDL cholesterol (Shargorodsky et al. 2010). Other trials have identified reductions to markers of DNA damage in lymphocytes (Ribeiro et al. 2007) and oxidative stress in red blood cells (Cheng et al. 2001). Improvements to biochemical cardiovascular markers, inflammation and oxidative stress represent potential mechanisms via which cognitive enhancements may occur. For this reason, it may be prudent to investigate these mechanisms of action alongside the cognitive effects of multivitamins.

The current randomised, double-blind, placebo-controlled trial therefore aimed to investigate both the cognitive and biochemical effects of 16 weeks treatment with a combined multivitamin, mineral and herbal supplement in healthy, community dwelling, elderly women. Recent evidence indicates that in the elderly, the cognitive processes most susceptible to cognitive deterioration, in turn, demonstrate the greatest improvements from nutraceutical intervention (Pipingas et al. 2008; Ryan et al. 2008). Cognitive domains prone to age-related deterioration include working memory, episodic memory and attention (Ronnlund et al. 2005). Conversely, other more declarative forms of memory including vocabulary and verbal IQ remain relatively intact in older adults (Christensen 2001), and may be less responsive to nutraceutical benefits. Consequently, in the current study, a validated battery of computerized, age-sensitive memory, attention and processing speed tasks (Pipingas et al. 2010) were utilized to test the hypothesis that the cognitive domains most vulnerable to the effects of age would benefit from multivitamin supplementation. Specifically it was predicted that improvements would be observed for speeded measures of memory and attention. As enhancements to verbal memory have also been identified following nutraceutical treatment (Summers et al. 2010), the California Verbal Learning Task II (CVLT-II) was included to assess treatment effects on immediate and delayed verbal memory.

The secondary aim of this study was to investigate biochemical changes associated with chronic intake of the multivitamin. Multivitamin supplementation has been demonstrated to exert beneficial effects on a range of biochemical and health parameters (Church et al. 2003; Earnest et al. 2003). In the current study, it was anticipated that the multivitamin would be capable of increasing blood vitamin levels, whilst lowering homocysteine and markers of inflammation and oxidative stress.

Methods

Study design

The trial was a 16-week, randomised, placebo-controlled, double-blind, parallel group investigation. Eligible participants were randomly allocated to receive the Swisse Women’s Ultivite 50 +™ supplement or a placebo with an allocation ratio of 1:1. This study was conducted according to the guidelines laid down in the Declaration of Helsinki. Approval for the trial was approved by the Swinburne Human Research Ethics committee.

Participants

Participants in the study were 56 community dwelling elderly females aged between 64 and 82 years. Participants were recruited from the community by way of advertisements which asked for women who were ‘concerned about their memory’, or ‘experiencing memory difficulties’, right handed and non smokers. Participants were required to answer ‘yes’ to the question ‘Do you feel like your memory is getting worse’? Exclusion criteria included a history of dementia, neurological disorder, stroke, epilepsy, Parkinson’s disease, mental illness, depression, anxiety disorders, head trauma or excessive alcohol use. Individuals using stimulant, anticholinergic or high dose anticoagulant medication were ineligible for participation. A 30-day wash-out period was required for participants taking a multivitamin supplement or any products containing G. biloba or St John’s Wort.

Screening

Potential participants were initially screened by telephone interview and subsequently attended a 1-h screening session during which written, informed consent was obtained. All testing was conducted at the Brain Sciences Institute, Swinburne University, Melbourne, Australia. The mini-mental state examination (MMSE) was administered to provide a brief indication of the participants cognitive status and to exclude any individuals obtaining a score (<24) possibly indicative of dementia (Folstein et al. 1975). A brief memory measure constructed by Jorm et al. (1997) was used to screen for subjective memory complaints. There were four items on this measure with scores ranging from 0 through to 8.

Participants underwent a medical examination with a medical practitioner to determine whether the individual was generally in good health, and to exclude any individuals using contraindicative medication.

Treatment

The treatment was a multivitamin, antioxidant and mineral formula with added herbal and antioxidant plant extracts (Swisse Women’s Ultivite 50 +™). Ingredients are shown in Table 1. The placebo tablets were identical in appearance to the multivitamin. Placebo tablets contained starch and 2 mg of Riboflavin (vitamin B2) to provide tablets with a similar smell and to produce a similar colouration of the urine. Supplements were packaged in blister packs containing seven tablets, labelled with each day of the week. Each participant received a box containing 17 sheets of supplements. Participants were instructed to take one supplement each day with breakfast for 16 weeks. An extra seven tablets were provided to enable participants to continue to supplement after the post-treatment appointment, until blood was collected.

Table 1 Ingredients of the multivitamin, mineral and herbal supplement

Randomisation and blinding

Randomisation was achieved using a computer generated random sampling set. The treatments were randomly allocated in blocks of four by the supplier; Swisse Vitamins Pty Ltd. Placebo and treatment were packaged in identical blister packs and contained in boxes numbered according to the randomisation schedule. Participants were allocated the next sequential number by the researcher involved in data collection at the baseline testing session. The blinding list was held by an investigator who was not involved in any aspect of data collection for the trial. Data was unblinded following preliminary data analysis.

Calculation of sample size

Due to the exploratory nature of this trial, a formal power analysis was not possible. Instead the sample size calculation was based on the findings of a previous trial which identified cognitive improvements with a moderate effect size, in 42 males on the same battery of cognitive tasks, following 5 weeks’ supplementation with a nutraceutical formula (Pipingas et al. 2008). In the current study the sample size was increased to 60 participants to increase the power of the study.

Procedure

Baseline testing

Participants attended the baseline testing session at 1030 or 1130 hours. Participants were requested to refrain from consuming tea or coffee for 2 h prior to their appointment. Verbal IQ was assessed at baseline using the contextual Aus-NART (Lucas et al. 2003). IQ was assessed to examine whether the treatment groups were matched on this variable at baseline. Participants completed the Swinburne University computerized cognitive assessment battery (SUCCAB) followed by CVLT-II. Participants were requested to attend a local pathology centre within 4 days to have blood collected. Blood was collected in the morning after a minimum 10-h fast. Multivitamin supplementation commenced after baseline blood collection.

Post-treatment testing

Participants returned for a post-treatment testing appointment, 16 weeks after baseline testing. Participants were instructed to refrain from taking the multivitamin on the morning of the post-treatment appointment. All subjects returned at the same time of the morning as at baseline. Subjects who could not attend on the scheduled return date were allowed a time window of 5 days to attend the testing session. The testing procedure from baseline was repeated. Alternate forms of the SUCCAB and CVLT-II were used. Participants followed the same procedure for blood sample collection.

Cognitive tests

Swinburne University computerised cognitive assessment battery

The SUCCAB was run in Pipscript04, a DOS based program which provides millisecond accuracy for stimulus presentation and response times. The SUCCAB subtests included in this trial have previously demonstrated good test–retest reliability, validity construct and sensitivity to age-related decline (Pipingas et al. 2010). The SUCCAB tasks were presented via computer (PC) and a hand-held button box was used to deliver all responses. Instructions were presented on the computer screen for each test and the investigator was available to answer any task-related questions. A practice trial was performed immediately prior to each task. The following tasks were performed by participants:

  • Simple Reaction Time—Participants responded to a white square, presented with a varying inter-stimulus interval.

  • Complex Reaction Time—Participants responded to a blue triangle or a red square, presented with a varying inter-stimulus interval, by pressing the corresponding coloured button.

  • Immediate and Delayed Recognition Memory—A series of abstract images were presented. A second series of images were then presented. Participants indicated which images they recognized from the previous block. Half the images were the same as the ones previously presented, and half were new. This process was repeated after 20 min with the other half of the images.

  • Stroop Congruent—The words ‘red’, ‘blue’, ‘green’ or ‘yellow’ were presented on the screen in corresponding coloured ink. Participants were required to respond to the word by pressing the same coloured button.

  • Stroop Incongruent—In this task, the word was different to the ink colour. Participants were required to ignore the written word and to respond to the colour of the ink by pressing the same coloured button.

  • Spatial Working Memory—Participants were required to remember the location of white squares in a 4 × 4 grid. In each trial, six grid locations were filled with a white square and participants were required to remember four locations.

    Contextual Recognition Memory—A series of pictures of everyday objects were presented at a location either at the top, bottom, right or left of the screen. A second series of the same images were then presented in centre of the screen. Participants responded by pressing the button which corresponded to the original location of the picture.

California Verbal Learning Task II

The CVLT-II is a word list learning task consisting of 16 items divided in to four semantic categories. Words were presented aurally to participants at a rate of 1.5 s. After each word list presentation participants repeated as many words as possible. This procedure was carried out five times. Immediate free recall was assessed after the presentation of a distracter list was presented. Delayed free recall was assessed after a delay of 25–30 min.

Biochemical analysis

Blood sample collection and analysis was performed by a pathology clinic with collection centres located around Melbourne. Homocysteine and vitamin B12 were measured in serum, vitamin B6 was determined from whole blood using high-performance liquid chromatography (HPLC) direct analysis of PLP (pyridoxal-5′-phosphate form). Serum vitamin E (tocopherol) was used as a measure of antioxidants. Inflammation was assessed using high sensitivity C-reactive protein (hCRP) and oxidative stress was measured using protein carbonyls. Protein Carbonyls proved a stable measure of oxidative stress and allow the detection of dose–response relationships (Collins 2005). The 2,4-dinitrophenylhydrazine (DNPH) reaction was used to measure protein carbonyl concentration in plasma. The quantity of protein-hydrozone produced during the DNPH reaction was quantified spectrophotometrically at an absorbance between 360 and 385 nm.

Urea and electrolytes were used as measures of kidney function. Elevated urea is an indicator of kidney function failure or dehydration. The electrolytes are the various salts in the bloodstream including sodium, potassium, chloride and bicarbonate. The Liver Function Test (LFT) was used to provide an indication of liver function and as a measure of active liver damage. Certain herbal preparations and botanicals have been associated with hepatotoxic effects (Stickel et al. 2000). Subsequently, in the current study the LFT was conducted to establish that the multivitamin and herbal formulation under investigation was safe and well tolerated by the elderly participants.

Statistical analysis

For all variables, the assumption of normality was assessed and variables skewed in a positive direction were subjected to a log X transformation, and the X 2 transformation was applied to variables skewed in a negative direction.

Cognitive data

The primary cognitive outcome measures were a memory and an attention composite measure. The memory measure consisted of averaged response times from the immediate, delayed and contextual recognition and working memory subtests and the attention composite measure comprised of the averaged response times from the reaction time and stroop tasks. As ceiling effects for accuracy were anticipated for simple and choice reaction time and stroop tasks (Pipingas et al. 2010), only response time was analysed for the composite measures. The composite measures were not corrected for multiple comparisons as these measures were shown to represent separate cognitive domains as confirmed by factor analysis conducted on baseline SUCCAB data. Individual cognitive subtests were only explored on the identification of a consistent pattern of change from baseline on all subtests of the composite measure.

For each subject, the change in response time from baseline to post-treatment was calculated. Univariate analysis of co-variance (ANCOVA) was used to examine the effect of treatment group on change in response time, with the relevant baseline score included as a covariate. On the identification of a significant treatment group effect, or a trend for this effect, post hoc t-tests were used to examine the difference between baseline and post-treatment measures for each group individually.

Biochemical data

Biochemical data was analysed using the same ANCOVA procedure as the cognitive data. Measures of blood nutrient levels including vitamins B12, B6 and E were adjusted for multiple comparisons using a Bonferroni correction. For these tests, a corrected p value of <0.02 was used to indicate statistical significance.

To determine whether changes in biochemical levels, attributed to the multivitamin, contributed to any cognitive changes, biochemical change from baseline values and a biochemical change × treatment group interaction were added as covariates to the ANCOVA for the relevant cognitive measure. The contribution of each biochemical measure was examined using separate ANCOVA models.

Results

Recruitment commenced in May 2008 and ceased in May 2009. The trial was ended once the planned 12 months of data collection were completed. Volunteers who completed the entire study were 51 of the 56 original participants. Participant demographics for the multivitamin and placebo group are shown in Table 2. One-way analysis of variance (ANOVA) indicated participant groups were matched on the demographic characteristics of age, verbal IQ, years of education, MMSE score, and subjective memory complaints score. In total, five participants from the multivitamin group and five from the placebo group were supplementing their diets daily with fish oil. All participants had been using fish oil supplements for a minimum of 6 months. At the time of screening, several participants reported using an individual vitamin supplement, with five participants using vitamin C, four participants using vitamin D, two participants using folate and two using vitamin B12. Participants were requested to discontinue individual vitamin supplementation, excluding fish oil, throughout the trial duration.

Table 2 Baseline participant demographics

Compliance

Remaining tablets were counted at the post-treatment appointment. Compliance as determined by the number of remaining tablets was shown to be high, with an average of 3 of the 112 supplements (97%) remaining at the end of the 16-week period. Participants with less than 80% compliance were excluded from the trial. One-way ANOVA indicated compliance did not differ between the multivitamin and placebo groups.

Treatment estimate

Of the 51 subjects, 39 were asked which treatment they believed they had received. A total of 18 participants were unsure which treatment they were allocated, six individuals correctly guessed they had received the multivitamin, eight correctly guessed the placebo and seven participants made an incorrect judgment.

Treatment side effects

One participant in the treatment group experienced nausea and vomiting after commencing multivitamin supplementation. One participant in the placebo group developed a mild rash during the supplementation period. Both subjects discontinued treatment and were withdrawn from the trial.

Excluded participants

Participant numbers at randomization and follow up are shown in Fig. 1. Participants with response times 3 or more standard deviations from the mean were excluded from the relevant cognitive analysis. Participants who were supplementing with vitamin B12 prior to commencing the trial were excluded from the relevant vitamin post-treatment analysis. Two participants were excluded from vitamin B6 analysis, and a further two participants from hCRP analysis as these values were over 2.5 standard deviations from the mean. Two participants were excluded from the LFT tests for high values above the normal range at both testing times. A further six participants were excluded from all post-treatment biochemical analysis for failing to attend the post-treatment blood collection appointment within 7 days of the post-treatment cognitive testing session.

Fig. 1
figure 1

Participant recruitment and treatment allocation flowchart

Cognitive results

The means and standard deviations for accuracy and response times on the SUCCAB at baseline and post-treatment testing are shown in Table 3. Figure 2 displays the changes in response time from the baseline to post-treatment session. One-way ANOVA revealed that that at baseline the multivitamin and placebo groups differed significantly on the memory composite measure (p < 0.01), but the attention measure.

Table 3 Means and standard deviations for cognitive assessments response times and accuracy
Fig. 2
figure 2

Changes in response time from baseline to post-treatment for information processing tasks: simple and complex reaction time, congruent and incongruent stroop; and memory tasks: contextual, immediate and delayed recognition and working memory from baseline to post-treatment

Primary outcomes

Examination of Fig. 2 demonstrates that response time decreases were larger for the multivitamin group than the placebo group for all memory tasks in the SUCCAB and changes in the SUCCAB subtests included in the attention composite measure were much smaller than the memory changes over the 4-month period. However, the effect of treatment group on the change in memory composite response time was not significant. Similarly, there was no effect of treatment group for the change in response time for the attention composite measure.

Secondary outcomes

SUCCAB memory subtests

As shown in Fig. 2, a greater reduction in memory response time was identified for all the memory subtests of the SUCCAB for the multivitamin treatment than the placebo. Subsequently, individual memory measures were examined for possible multivitamin treatment effects. A significant effect of treatment group was identified for the spatial working memory task (F 1,47 = 4.09, p = 0.049, η 2 = 0.09). Post hoc t-tests revealed there was a decrease in response time from baseline to post-treatment for the multivitamin group (p = 0.025) and a non significant increase for the placebo group (p = 0.32). There were no significant effects of treatment group for the immediate, contextual or delayed recognition memory tasks.

Verbal memory

There were no significant treatment group effects identified for the CVLT-II measures.

Biochemical results

Means and standard deviations for the baseline and post-treatment assessment are shown in Table 4 for biochemical variables. A series of one-way ANOVAs were conducted to investigate group differences at baseline, with vitamin B6 shown to be significant at the p < 0.01 level. No other biochemical variables differed significantly at baseline.

Table 4 Means and standard deviations for biochemical blood measures from baseline and post-treatment sessions

Homocysteine and B vitamins

As shown in Table 4, homocysteine was reduced in the multivitamin group, but not the placebo group and this treatment effect was significant (F 1,41 = 4.81), p =0.03, η 2 = 0.11). Post hoc t-tests indicated there was a trend for homocysteine to be reduced from baseline to post-treatment for the multivitamin treatment (p = 0.05), but not the placebo (p =0.67). A significant treatment group effect was identified for vitamin B12 levels (F 1,39 = 25.1), p <0.001, η 2 = 0.39). The increase in levels of vitamin B12 from baseline to post treatment testing for the multivitamin was significant (p < 0.001), whilst there was no significant change for the placebo group (p = 0.13). The results for vitamin B6 levels also revealed that there was a significant treatment group effect (F 1,36 = 43.91, p < 0.001, η 2 = 0.55). Post hoc t-tests indicated that there was a significant increase in levels of vitamin B6 from baseline to post treatment testing (p < 0.001), whilst there was no significant change for the placebo group (p = 0.33).

Vitamin E

The results for vitamin E demonstrate that there was a trend for an effect of treatment group (F 1,41 =4.21, p = 0.047, η 2 = 0.09. Post hoc t-tests revealed the change from baseline to post-treatment vitamin E levels did not reach statistical significance for the multivitamin (p = 0.07) or placebo groups (p = 0.08).

Fibrinogen, inflammation and oxidative stress

There were no significant effects of treatment group for fibrinogen or the hsCRP marker of inflammation.

Blood safety measures

The multivitamin treatment was not found to influence blood safety parameters. There were no significant effects of treatment group for the electrolytes, urea or liver function tests.

Biochemical contributions to cognitive change

ANCOVA revealed that the treatment group effect for spatial working memory remained significant when controlling for change in levels of vitamin B6 (F 1,35 = 6.34, p = 0.02, η 2 = 0.15) and vitamin E (F 1,38 = 4.36, p = 0.04, η 2 = 0.10). Examination of the covariates for each vitamin revealed the changes in these vitamin levels and the biochemical change × treatment group interactions were not significantly related to the working memory changes. The treatment group effect for spatial working memory was no longer significant when controlling for changes in levels of homocysteine or vitamin B12; however, examination of the covariates also demonstrated no relationship between changes in biochemical levels or the biochemical change × treatment group interaction with change in cognition.

Discussion

The current study aimed to investigate the effects of 16 weeks supplementation of a multivitamin and herbal formula on cognition in community dwelling, elderly women. In line with the hypothesis that the multivitamin treatment would impart cognitive benefits to the domains most vulnerable to the effects of age, the results revealed there was a significant reduction in response time for spatial working memory. However, there were no observed treatment effects for corresponding measures of attention, or verbal memory using the CVLT-II. In terms of biochemical changes, multivitamin supplementation decreased homocysteine, increased vitamins B6 and B12, with a trend for vitamin E to increase. Markers of inflammation and oxidative stress were not improved by the multivitamin. Haematological blood safety parameters indicated the multivitamin was safe and well tolerated.

In this study, multivitamin supplementation improved spatial working memory performance. Working memory is particularly vulnerable to the effects of ageing, and age-related detriments to performance have been observed in a cross-sectional lifespan investigation using the same measure of spatial working memory as the current study (Pipingas et al. 2010). Participants in the current study were females who reported subjective memory complaints, a condition which may relate to early manifestations of memory impairment (Jorm et al. 2001). Whilst it is not known whether the same memory effects would be observed in a comparable group of males, it has been suggested that amongst the elderly there are subgroups that will experience the greatest cognitive benefits from dietary interventions (Balk et al. 2007). By focussing on a participant group experiencing below optimum cognitive functioning, combined with age-sensitive cognitive measures, the current study may have been more capable of identifying cognitive improvements with multivitamin supplementation than trials which identified negative treatment effects (Wolters et al. 2005; McNeill et al. 2007).

Findings of improved cognitive performance are consistent with the identification of multivitamin treatment effects on computerized measures of fluid intelligence in younger participant samples (Haskell et al. 2010; Kennedy et al. 2010). The most robust cognitive benefits were specific to the domain of working memory, providing support to the predictions of other researchers that measures of fluid intelligence may be responsive to treatment with vitamin formulations due to the role of these nutrients in the maintenance of CNS function and production of neurotransmitters relevant to memory function (Cockle et al. 2000; Bryan et al. 2002; Haskell et al. 2008). Working memory represents an important component of fluid intelligence and growing evidence indicates that working memory may represent a cognitive domain which benefits preferentially from nutraceutical intervention (Pipingas et al. 2008; Ryan et al. 2008; MacReady et al. 2011). Performance on the same spatial working memory task included in the current study has demonstrated improvements following 5 weeks supplementation with a flavonoid blend (Pipingas et al. 2008), indicating that this measure is sensitive to both multivitamin and flavonoid intervention.

Mechanisms of improvement

Whilst findings of the current study indicated the multivitamin increased levels of vitamins B6, B12, and E, and reduced homocysteine, alterations in these biochemical parameters were not related to the improvements to working memory response time. These results may indicate that cognitive enhancements occurred either as a result of additive effects of the vitamin, mineral and herbal components, or through components of the multivitamin which were not measured. Folate and B12 are required for the production of SAM, which in turn contributes to the production of myelin, neurotransmitters, and membrane phospholipids (Rosenberg and Miller 1992). It is conceivable that by increasing these nutrients via dietary supplementation, there may be a beneficial effect on the function and integrity of the nervous system. Evidence for this premise has been obtained from a trial conducted by Kennedy et al. (2010) which revealed that in younger adults, 1 months supplementation with a vitamin B complex, vitamin C and mineral formula improved performance on a mental calculation task designed to tax the processes of psychomotor function, attention, working memory and executive function. However, if a generalized improvement to the nervous system was to solely account for the improvements to speed of memory response in the current study, similar effects would be anticipated for the measures of processing speed and attention. The results did not reveal improvements to these measures.

The multivitamin decreased homocysteine by 1.2 μmol/l, the same figure previously reported after 12 weeks multivitamin supplementation (Earnest et al. 2002) and slightly lower than the 1.57 μmol/l reduction reported by Summers et al. (2010) after 16 weeks. Only one trial has provided convincing evidence that reducing homocysteine through folate supplementation can improve memory in healthy elderly, and this trial was of 3 years’ duration (Durga et al. 2007). Elevated homocysteine has been related to hippocampal atrophy (Williams et al. 2002; Den Heijer et al. 2003) and more distributed brain atrophy (Rajagopalan et al. 2011). Lowering homocysteine through 12 months B vitamin treatment has slowed the rate of brain atrophy in subjects with mild cognitive impairment (Smith et al. 2010). In the present study, lowering homocysteine through multivitamin supplementation did not appear to influence cognition. Reduction of homocysteine may need to occur over a longer time period before brain structural parameters and cognition are influenced.

In the current study, there was a trend for levels of vitamin E to increase following multivitamin supplementation, and to decrease in those allocated to the placebo. In addition to lowering oxidative stress, it has been suggested that fat soluble antioxidants such as vitamin E may also exert direct effects on the brain (Sen and Khanna 2010). In rodents, vitamin E prevents ischemia-induced neuronal death in hippocampal cells (Hara et al. 1990) and has been shown to facilitate LTP in neurons (Xie and Sastry 1993). Vitamin E interacts synergistically with selenium (Bourre 2006) and other plant derived antioxidants, to increase overall antioxidant capacity (Fuhrman et al. 2000). Although increases to levels of this vitamin alone were not related to cognitive changes in this study, it is still possible that additive actions of vitamin E with other components of the multivitamin contributed to cognitive changes.

Despite multivitamin treatment-related increases to levels of vitamin E, there were no corresponding decreases to markers of oxidative stress or inflammation. Similarly, in a study of middle aged men there were no benefits of 8 weeks B vitamin, antioxidant or combined B vitamin and antioxidant treatment on C-reactive protein (O'Doherty et al. 2010). In contrast, reductions to C-reactive protein have been demonstrated following a longer period of 6 months multivitamin supplementation in subjects aged 30–70 years. Prior intervention trials have also shown that 4 weeks multivitamin treatment reduced levels of DNA oxidative stress in elderly subjects (Ribeiro et al. 2007) and 12 weeks supplementation with combined vitamins C and E attenuated urinary isoprostanes in individuals at risk of cognitive decline (Clarke et al. 2003). However, other intervention studies have indicated that benefits may be restricted to individuals with low baseline antioxidant status (Carty et al. 2000) or elevated oxidative stress (Block et al. 2008). At baseline, levels of protein carbonyls in the present study were comparable to those previously measured in healthy elderly women (Kasapoglu and Özben 2001), and may not represent an elevated state of oxidative stress. These findings suggest that biochemical actions of multivitamin supplementation should not be inferred from other trials using different supplements or subject groups.

Selected memory improvements of the multivitamin may also be due to the botanicals G. biloba or Bacopa monierra which have demonstrated cognitive enhancing effects in humans, especially in the domains of learning and memory (Stough et al. 2001; Mix and Crews 2002). Numerous nootropic mechanisms of G. biloba have been proposed, including antioxidant effects, which inhibit neuronal apoptosis (Ahlemeyer and Krieglstein 2003) and increased cerebral blood flow due to vasodilation (Mantle et al. 2000). Actions of G. biloba may also involve direct effects on the cholinergic neurotransmitter system essential for the regulation of cognition (Di Renzo 2000). Cognitive enhancing mechanisms of bacopa are thought to include increasing antioxidant activity in frontal cortical, striatal and hippocampal brain regions (Bhattacharya et al. 2000) and reducing amyloid in the brain (Holcomb et al. 2006). Combined, the influence of these plant extracts on brain function may have contributed to observed memory enhancements of the multivitamin treatment.

Other plant derived flavonoid extracts may also have exerted cognitive improvements as flavonoid extracts have previously been demonstrated to improve non-verbal working memory, the same cognitive domain affected by the multivitamin in the present study (Pipingas et al. 2008; Ryan et al. 2008). Based on the results of rodent investigations, specific mechanisms in the hippocampus have been proposed, including reduced cell death by increasing intracellular glutathione, reduced reactive oxygen species and prevention of calcium influx which can be toxic to neurons (Ishige et al. 2001). It has been suggested that the beneficial effects of flavonoids on human cognitive performance may be due to interactions with neuronal signalling pathways to promote LTP and synaptic plasticity (Spencer 2008), or peripheral and cerebral vascular actions, which contribute to growth of new neurons in the hippocampus (Spencer 2009). Subsequently, it is possible that the memory benefits identified in the current study could have been due to mechanisms of flavonoid ingredients or the synergistic effects of the flavonoids with other vitamin and mineral components.

This trial is not without limitations and the absence of participant dietary information at baseline and over the supplementation period represents a potential constraint of this study. Findings that blood nutrient levels increased for the multivitamin group from baseline to post-treatment, whilst there were no statistical changes in biochemical levels for the placebo group, provide assurance that increases in blood vitamin levels were due to the multivitamin treatment and not to extraneous dietary factors. Future studies may benefit from standardising dietary intake on the assessment days to further reduce any variability due to the potentially confounding influence of diet.

Although not reaching statistical significance, greater improvements for the multivitamin treatment were observed on the memory composite measure and all memory subtests of the cognitive battery. Practice effects were also observed on these measures and may have precluded the identification of any cognitive benefits due multivitamin supplementation. The inclusion of a separate practice session prior to baseline testing may be necessary to eliminate learning effects in future trials. In a recent investigation it was revealed that a high level of cognitive demand was necessary to capture nutraceutical effects of a cognitive enhancing polyphenol (Scholey et al. 2010). The addition of cognitive tasks of higher or graded difficulty may help to determine whether the cognitive enhancing effects of multivitamins in the elderly are specific to working memory or are instead related to a wider range of cognitive processes.

Summary and conclusion

This study demonstrated that 16 weeks multivitamin supplementation was capable of improving speed of spatial working memory response in community dwelling elderly women. To confirm these results, replication of this trial in a larger sample including elderly males is advisable. Multivitamin supplementation decreased levels of homocysteine and increased levels of vitamin B6 and B12, with a trend for vitamin E to increase, however these individual components did not appear to contribute to the observed working memory enhancements. Future trials should examine alternate mechanisms of cognitive enhancement including a broader range of biochemical and cardiovascular parameters.