Psychological well-being is high on the list of public health policy priorities, with the emphasis on prevention(Reference Wanless1). Zinc is an essential trace element(Reference Hambidge2) that is present in large quantities in the brain and may be important for the maintenance of psychological well-being. Zinc-containing neurons are found in the forebrain interconnecting the cerebral cortex (the cognitive or ‘thinking’ part of the brain) and the limbic system (the affective or ‘feeling’ part of the brain)(Reference Frederickson, Won Suh and Silva3). The synaptic vesicles of the hippocampus and amygdala, which are both limbic structures, contain particularly large quantities of zinc(Reference Colvin, Davis and Nipper4), implying a role for zinc in the control of emotion. Zinc-containing neurons are receptive to glutamine, an amino acid neurotransmitter that is excitatory and that exerts a neuromodulatory effect on post-synaptic glutamate receptors within the hypothalamic–pituitary–adrenal axis (HPA) system(Reference Colvin, Davis and Nipper4, Reference Szewczyk, Poleszak and Wlaz5). Rodent studies have indicated that this action occurs mainly through the hippocampus synapses, which are involved in the regulation of the HPA system(Reference Matias, Saggau and Quinta-Ferreira6–Reference Takeda11). Zinc may also be involved in the control of the serotonergic system(Reference Szewczyk, Poleszak and Wlaz5, Reference Cichy, Sowa-Kucma and Legutko12, Reference Barrondo and Salles13). Zinc has been shown to increase the density of 5-HT serotonin receptors in the hippocampus and frontal cortex in rats(Reference Cichy, Sowa-Kucma and Legutko12). Post-mortem evidence suggests that zinc may have a role in the nutritional methylation processes(Reference Michel, Frangou and Thiemeyer14) that are also believed to be important in the regulation of mood(Reference Kaplan, Field and Crawford15). The human central nervous system appears to be adversely affected by zinc deficiency. Lower serum zinc is associated with depression(Reference Marcellini, Giuli and Papa16–Reference Maes, Vandoolaeghe and Neels20) and a marker of treatment resistance in clinical depression(Reference Maes, Vandoolaeghe and Neels20, Reference Siwek, Dudek and Paul21). Lower serum zinc is also associated with disordered behaviour(Reference Walsh, Isaacson and Rehman22–Reference Hanson, Malecha and Mackenzie24). Zinc supplementation has also been shown to ameliorate depression in clinical groups(Reference Maes, Neels and Demedts25, Reference Levenson26). There is an apparent dearth of human studies of zinc and well-being in healthy (non-clinical) groups. Among the few studies of healthy individuals is a recent cross-sectional study of Korean women (n 570), which indicated that among other nutrients and trace elements, lower dietary zinc intake was associated with higher stress scores on the short version of the Psychological Well-being Index(Reference Hwang, Lee and Kim27). A recent double-blind, placebo-controlled trial of zinc (7 mg/d) in young women (n 30) found reduced scores on the anger/hostility and depression scales of the Profile of Mood States(Reference Sawada and Yokoi28). These studies imply a link between zinc and psychological well-being in health. There appear to be no existing trials of zinc and well-being or mood in men and/or older people.
Europe and much of the industrialised Western world has a growing ageing population(Reference Wanless1), hence, the need to maintain psychological well-being and quality of life in this vulnerable age group. The elderly are particularly at risk of zinc deficiency, the likelihood of which increases with age(Reference Seiler29–Reference Maes, de Vos and Wauters33) as a result of a range of physiological, social, psychological and economic factors(Reference Reichies, Felsenberg and Gessner34) that may be associated with the decreased consumption of zinc-rich foods(Reference Drewnowski and Shulz35–Reference Webb, Schofield and Lazerus38). Dysregulation of the HPA system is associated with the increased risk of CVD, diabetes, cancer, inflammatory conditions and neurodegenerative disorders(Reference Cole and Dendukuri39). HPA system activity has been shown previously to be related to positive affect (PA) in the elderly(Reference Simpson, McConville and Rae40). Maintaining a higher PA in old age, therefore, is potentially important for general health and well-being as well as disease prevention(Reference Ostir, Ottenbacher and Markides41) in otherwise healthy older people. Assuming that zinc exerts a neuromodulatory effect through the HPA system, optimising zinc status in older people, therefore, is likely to have far-reaching benefits to psychological health. Yet, there do not appear to be any previous studies that have considered zinc and affect in healthy older individuals. Existing evidence for benefits of zinc to psychological well-being is mainly derived from post hoc investigation using psychiatric in-patients. Trials in healthy populations are necessary to further our understanding of how dietary constituents such as zinc impact upon health and well-being and to provide evidence to support or refute potential functional claims for zinc-rich food products. Trials in healthy individuals are especially important, given the widespread and increasing use of self-prescribed nutrient supplements by the older public(Reference Goh, Vitry and Semple42). The Zenith Study aimed to assess the effects of zinc supplementation in the normal healthy older population. It was necessary, therefore, to intervene with zinc at dosages that would be available for purchase over the counter without prescription. This randomised, placebo-controlled, double-blind intervention study reports the effect upon PA and negative affect (NA) of zinc (15mg/d and 30 mg/d) supplements administered orally over 6 months compared to placebo in healthy 55–70-year-olds recruited to the Zenith Study in Coleraine and Clermont-Ferrand and ≥70-year-olds in Grenoble and Rome. This research meets an imperative for controlled supplementation trials investigating the effect of zinc on psychological well-being in the older population. It is hypothesised that zinc supplementation will alter subjective affective state compared to placebo.
Methods
The Zenith Study was a randomised, double-blind, placebo-controlled intervention trial of zinc in older adults, which was conducted in four European centres. Ethics approval for the study was obtained from recognised research centres in each country. Informed, written consent was obtained from each participant.
Sampling
Volunteers were outreached in different ways including through local television and radio broadcast, posters and through leaflets distributed in supermarkets, clubs that free-living elderly people frequently visit, health centres, and family doctors. Prospective volunteers then initiated contact with the research group who then arranged an appointment for the preliminary screening. The volunteers were screened first by clinical examination, anthropometry, smoking and alcohol consumption, cognitive impairment, depression and biochemistry profiles. Those selected for inclusion in the study were apparently healthy, not morbidly obese and with a BMI of between 20 kg/m2 and 30 kg/m2(43). Only those who had no evidence of dementia, as indicated by a score >23 in the Mini Mental State Examination (MMSE) test(Reference Folstein, Folstein and McHugh44), were included. Volunteers were screened for depression using the Geriatric Depression Scale (GDS)(Reference Yesavage, Brink and Rose45). The GDS has been validated as a screening tool for depression against the DSM-IV criteria according to which only those scoring ≥6 can be considered not clinically depressed(Reference Almeida and Almeida46). Those with a score of ≤5 were included. Participants were also excluded if they had a positive serology for HIV or hepatitis C, if they smoked more than ten cigarettes per day; consumed alcohol >30 g/d for men and >20 g/d for women; had unusual dietary habits (vegetarianism and veganism); used a mineral supplement in the 3 months preceding the study; used more than three drugs daily if aged 55–70 years; used more than four drugs daily if over 70 years; used drugs including antidepressants, laxatives and hormone replacement therapy; or had a pathological disease (cancer, diabetes, insufficient renal or hepatic performance, malabsorption and inflammatory chronic pathologies). Complete details about screening and recruitment processes have been described previously by Polito et al.(Reference Polito, Meunier and Andriollo-Sanchez47).
Serum zinc
Unfortunately, there is no single accurate measure of zinc status(Reference Gibson48, Reference Miller, Krebs and Hambridge49). For the purpose of the present study, serum zinc was taken as a putative indicator of current zinc status(Reference Anriollo-Sanchez, Hininger-Favier and Meunier50). Fasting blood samples were obtained for the zinc assay in the morning. Blood samples for the zinc assay were collected via venepuncture by qualified the phlebotomists, using trace element-free Vacutainer® tubes (Becton, Dickinson and company Ltd, Oxford, UK). The blood sample was taken in the morning after a 12 h fast by the participants. Immediately after drawing, the blood samples were kept on ice, and then centrifuged at 1000g for 15 min at 4°C. Serum zinc was immediately isolated, aliquoted and stored at −80°C. Serum zinc concentrations at baseline were determined at INRA Clermont-Ferrand, France, by flame atomic absorption spectrometry using a Perkin-Elmer 560 instrument (Perkin Elmer, Cambridge, UK)(Reference Anriollo-Sanchez, Hininger-Favier and Meunier50). Internal quality control was checked using Seronorm® trace element serum (Sero®, Billingstad, Norway).
Dietary zinc
Dietary zinc intake was assessed at baseline by means of a semi-structured standardised 4 d food diary over two weekdays and two weekend days consecutively, which included estimates of portion size, the brand name of the product and/or the recipe if cooked from fresh. The information in the food diaries was analysed using the NetWISP version 3·0 (Tinuviel Software Anglesey, UK) database. A full account of the dietary assessment procedure has been provided previously by Polito et al.(Reference Polito, Intorre and Andriollo-Sanchez51).
Positive and Negative Affect Scale
The Positive and Negative Affect Schedule (PANAS) is a well-validated twenty-item self-reported psychometric measure of subjective mood developed by Watson et al.(Reference Watson, Clarke and Tellegen52). Watson and Tellegen(Reference Watson and Tellegen53) proposed that there are two distinct dimensions of mood state, PA and NA, which together account for over two-thirds of the variance in mood. PA is associated with feelings of alertness, enthusiasm and happiness and NA with displeasure and dissatisfaction(Reference Watson and Clark54). PANAS scores have been shown to correlate with salivary cortisol levels(Reference Simpson, McConville and Rae40) and whole blood serotonin levels in healthy individuals(Reference Williams, Stewart-Knox and Helander55, Reference Duffy, Stewart-Knox and McConville56). The scales have been found to have high internal consistency, with Cronbach’s α ranging from 0·84 to 0·87 for the NA scale and 0·84 to 0·90 for the PA scale(Reference Watson and Walker57). Repeated measures of affect were taken during the week before the centre visit, including one further measure taken on the morning of testing. The participants were asked to complete four self-reported PANAS questionnaires daily (on rising, after lunch, after dinner, and before going to bed) for four consecutive days at baseline, 3 and 6 months. The scale took approximately 2 min to complete. The items were: interested; distressed; excited; upset; strong; guilty; scared; hostile; enthusiastic; proud; irritable; alert; ashamed; inspired; nervous; determined; attentive; jittery; active; and afraid. Responses to the PANAS were recorded on a 5-point Likert scale ranging from ‘not at all’ = 1 to ‘extremely’ = 5. A score for each scale was obtained by summing item scores.
Intervention
Eligible volunteers were randomly assigned to three groups to receive either 15 mg/d or 30 mg/d zinc gluconate or placebo orally. Zinc capsules were issued in dated pillboxes at baseline and at 3 and 6 months into the supplementation phase. The participants were instructed to swallow the capsules with breakfast. The degree of compliance was estimated by collecting the number of returned capsules at 3 and 6 months. Compliance was tested and was found to be >98 % for all groups.
Data analysis
The mood data were analysed using a mixed ANOVA. Separate analyses were computed for the 55–70-year-olds (n 188; from Coleraine and Clermont-Ferrand) and for the ≥70-year-olds (n 199; from Grenoble and Rome). There were three between-group factors (sex, centre and treatment) and one repeated measure (baseline, 3 and 6 months post-intervention) included in each ANOVA. Mean PA and NA were treated as separate dependent variables, which also required separate ANOVAs. In the absence of relevant previous studies in this area, a conservative value of f 2 = 0·025 for effect sizes was adopted. For the mood scale it was assumed that the correlations among each of the repeated measures were 0·42 (taken from the PANAS manual). Power calculations were performed using GPOWER general power analysis computer program (G*Power, Bonn, Germany)(Reference Erdfelder, Faul and Buchner58). With α = 0·05 and assuming 180 participants within each age group (55–70 and ≥70 years), the power values for the between-groups and within-groups and interaction effects were 0·36, 0·99 and 0·81, respectively. As the interaction effects (particularly the treatment–time interaction) were of prime concern in the present study, the sample sizes for both the younger (n 188) and the older (n 199) age groups were regarded as satisfactory. Significance was set at P < 0·05. The Statistical Package for Social Sciences statistical software package version 11·5 (SPSS UK Ltd., Feltham, UK) was used for statistical analysis.
Results
Approximately 10–15 % of those who initiated contact were included in the study. Equal proportions of healthy male and female older volunteers were recruited from four European centres. Coleraine, Northern Ireland (n 93) and Clermont-Ferrand, France (n 95) recruited participants aged 55–70 years, while Rome, Italy (n 108) and Grenoble, France (n 91) recruited participants who were >70 years of age. In total, 387 participants successfully completed the trial (Table 1). A full description of the sample characteristics has been provided elsewhere by Simpson et al.(Reference Simpson, O’Connor and Livingstone59).
Table 1 Sociodemographic variables for each region, given as percentages
Total participants=387: Northern Ireland = 93; Clermont-Ferrand = 95; Rome = 108; and Grenoble = 91.
Baseline measures
Neither sex nor age differed between the treatment groups. The younger (55–70 years) age group comprised a higher proportion (84 %) of professional occupations than the older (≥70 years) age group (68 %). A higher proportion of those in the younger age group (96 %) than the older age group (35 %) were educated to the tertiary level. Neither social class nor educational level differed between the treatment groups. There were no apparent differences by age group or treatment group in dietary zinc intake at baseline. There were no differences in MMSE or (GDS) scores by age group or treatment condition(Reference Maylor, Simpson and Secker60).
Serum zinc
Mean serum zinc was within the normal range (11–18 μmol/l)(Reference McConville, Simpson and Rae61) for the placebo (mean = 13·20 (sd 1·69) μmol/l), 15 mg (mean = 13·28 (sd 1·84) μmol/l) and 30 mg (mean = 13·13 (sd 1·63) μmol/l) supplemented groups at baseline and remained so throughout the intervention period. Serum zinc increased over time (F(4197) = 11·021, P = 0·000) in both the 15 mg (mean = 13·99 (sd 2·47) μmol/l; P = 0·018) and 30 mg (mean = 15·03 (sd 3·17) μmol/l; P = 0·000) supplemented groups compared to placebo (mean = 13·05 (sd 1·66) μmol/l) suggesting compliance with the intervention. Serum zinc concentrations in response to zinc (30 mg) were higher among those recruited in Rome (mean = 16·26 (sd 3·41) μmol/l) than Grenoble (mean = 13·64 (sd 2·21) μmol/l) post-intervention (F(4197) = 3·526, P = 0·008).
Affect (the Positive and Negative Affect Scale) and zinc
There were no treatment (15 mg/d zinc; 30 mg/d zinc; or placebo) × time (baseline; 3 months; or 6 months post-intervention) or interaction effects, indicating that the treatment had no differential effects on affect (PANAS) over time. For NA among the 55–70-year-olds, the interaction effect was F(4340) = 1·56, P = 0·185 (Table 1) and for the ≥70-year-olds, F(4372) = 0·466, P = 0·761 (Table 2). Interaction effects for PA among the 55–70-year-olds were F(4340) = 0·37, P = 0·833 (Table 2) and among the ≥70-year-olds were F(4372) = 1·32, P = 0·261 (Table 3). None of the triple interaction effects involving treatment and time were significant.
Table 2 PANAS scores for each treatment group (55–70-year-olds/Coleraine and Clermont-Ferrand) over time (n 188)
PANAS, Positive and Negative Affect Score.
Table 3 PANAS scores for each treatment group (≥70-year-olds/Rome and Grenoble) over time (n 199)
PANAS, Positive and Negative Affect Score.
Centre and affect (the Positive and Negative Affect Scale)
There was a significant treatment × centre interaction effect for NA among the ≥70-year age group (F(1186) = 3·73, P = 0·025). For the placebo, 15 mg/d and 30 mg/d treatment groups, the means were 11·45, 13·43 and 11·94, respectively, for Rome and 13·06, 12·83 and 14·38, respectively, for Grenoble. For Grenoble, the trend for NA was U-shaped showing a decrease at 3 months into the intervention and returning to baseline at 6 months. For Rome, the trend for NA was an inverted U-shape such that NA was increased at 3 months into the intervention and returning to baseline at 6 months. There were no significant interactions between treatment and centre for PA in either the 55–70-year or ≥70-year age group.
Gender and affect (the Positive and Negative Affect Scale)
With the exception of a significant gender × centre interaction effect for NA among the 55–70-year-olds (F(1170) = 4·71, P = 0·031), there were no significant main effects or interaction effects for gender. The significant gender × centre interaction effect was such that male volunteers reported higher NA than female volunteers in the Coleraine sample (mean = 12·00 and 11·39, respectively), whereas the reverse was true in the Clermont-Ferrand sample (mean = 11·48 and 12·14).
Discussion
Studies of clinically depressed(Reference Bodnor and Wisner18–Reference Hanson, Malecha and Mackenzie24) and healthy(Reference Hwang, Lee and Kim27, Reference Sawada and Yokoi28) groups have suggested a link between zinc status and psychological well-being. Given evidence to suggest that for a multiplicity of reasons, ageing can be associated with zinc depletion(Reference Seiler, Itin and Stahelin31), it was hypothesised that supplemented zinc would enhance mood in older people. It was, therefore, surprising to find no alteration in PA or NA in response to zinc (15 mg/d or 30 mg/d) compared to placebo in either the younger (55–70 years) or the older (≥70 years) age group, who had lower PA than the younger age group at baseline(Reference McConville, Simpson and Rae61). Both NA and PA were unaffected by zinc supplementation in either age group. In the older age group (≥70 years), NA increased among those in Rome and decreased among those in Grenoble 3 months into the intervention and returned to baseline in both groups at 6 months. This trend is difficult to explain. Although serum zinc was higher among those in Rome than Grenoble at 3 months, serum zinc is not an accurate indicator of zinc status and could have been altered in response to a range of factors unrelated to affect(Reference Miller, Krebs and Hambridge49). Increased serum zinc concentrations post-treatment, however, suggest good compliance with the intervention in both groups. The PANAS is a well-validated subjective (self-reported) psychometric measure of affect(Reference Watson, Clarke and Tellegen52), which has been widely used and previously used in other studies involving elderly adults(Reference Hill, Van Boxtel and Ponds62). The PANAS has also been shown to correlate with the biochemical markers of affect(Reference Simpson, McConville and Rae40, Reference Williams, Stewart-Knox and Helander55, Reference Duffy, Stewart-Knox and McConville56). This correlation indicates that although the scales measure subjective affect, the data generated can be considered sensitive to corresponding changes in neurochemical activity. The PANAS can, therefore, be assumed to have good concurrent validity for the measurement of affect. Although the PA data collected for the present study agree with those of the established norms, the NA scores appear to be less than those suggested by established norms for older people(Reference Watson, Clarke and Tellegen52). The relatively low NA reported by our sample at baseline may explain the lack of change in NA following zinc supplementation. The apparent lack of change in affect in response to zinc could also be because the participants were zinc replete(Reference Anriollo-Sanchez, Hininger-Favier and Meunier50). Screening procedures were stringent and care was taken to recruit only physically and psychologically healthy older people to the Zenith Study. Health and well-being tend to be confounded with social factors. It is possible that in selecting only healthy individuals, those from socio-economically deprived segments of society have been excluded. That our sample was biased towards the higher social classes and those who had spent a relatively longer period in education (Table 1) may also explain the apparently good psychological well-being among our sample and lack of response to the intervention. The participants, by virtue of having volunteered, may not be entirely representative of the general population, potentially limiting the degree to which the findings can be extrapolated to the general older population. Those who have spent longer periods of time in education and those in more privileged social classes tend to have lower mortality and morbidity(Reference Wanless1) and may, therefore, also have a greater chance of being recruited to such studies.
Conclusion
The present study seems to be the first randomised, double-blind, placebo-controlled intervention trial investigating the impact of supplemented zinc upon affect in healthy older adults. Intervention outcomes were assessed using the PANAS, which is a well-validated subjective (self-reported) psychometric measure of affect. The study was also strong in that it used a relatively large sample comprising late middle-aged and elderly individuals from both northern and southern European populations with adequate power with which to uphold the null hypothesis and within which sex and cultural effects were controlled. These data suggest that zinc supplementation at considered 15 mg/d or 30 mg/d does not alter mood in healthy elderly European adults.
Acknowledgements
The project was funded by the European Commission ‘Quality of Life and Management of Living Resources’ Fifth Framework Programme, Contract no. QLK1-CT-2001-00168. The authors have no conflicts of interest. B.J.S.-K. contributed to the design of the study and drafted the paper; G.R. undertook the statistical analysis; E.E.A.S. undertook data collection and database development; J.M.O. coordinated research at University of Ulster; C.Mc.C. contributed to the mood protocol; A.P. supervised the data collection at INRAN; M.A.-S. supervised the data collection at Grenoble; C.C. supervised the data collection at INRA and coordinated Zenith; and J.J.S. contributed to the nutritional protocol and advised on the zinc intervention.
