Effect of multivitamin and multimineral supplements on morbidity from infections in older people (MAVIS trial): pragmatic, randomised, double blind, placebo controlled trialBMJ 2005; 331 doi: https://doi.org/10.1136/bmj.331.7512.324 (Published 04 August 2005) Cite this as: BMJ 2005;331:324
- Alison Avenell, clinical research fellow ()1,
- Marion K Campbell, deputy director1,
- Jonathan A Cook, statistician1,
- Philip C Hannaford, NHS Grampian professor of primary care2,
- Mary M Kilonzo, research fellow3,
- Geraldine McNeill, research fellow4,
- Anne C Milne, research fellow1,
- Craig R Ramsay, senior statistician1,
- D Gwyn Seymour, professor of medicine for the elderly5,
- Audrey I Stephen1, research assistant,
- Luke D Vale3, senior research fellow
- 1 Health Services Research Unit, School of Medicine, University of Aberdeen, Foresterhill, Aberdeen AB25 2ZD
- 2 Department of General Practice and Primary Care, School of Medicine, University of Aberdeen, Foresterhill Health Centre, Aberdeen AB25 2AY
- 3 Health Economics Research Unit, School of Medicine, University of Aberdeen, Foresterhill, Aberdeen AB25 2ZD
- 4 Department of Environmental and Occupational Medicine, School of Medicine, University of Aberdeen, Liberty Safe Work Research Centre, Aberdeen AB25 2ZP
- 5 Department of Medicine and Therapeutics, School of Medicine, University of Aberdeen, Aberdeen AB25 2ZD
- Correspondence to: A Avenell
- Accepted 9 June 2005
Objective To examine whether supplementation with multivitamins and multiminerals influences self reported days of infection, use of health services, and quality of life in people aged 65 or over.
Design Randomised, placebo controlled trial, with blinding of participants, outcome assessors, and investigators.
Setting Communities associated with six general practices in Grampian, Scotland.
Participants 910 men and women aged 65 or over who did not take vitamins or minerals.
Interventions Daily multivitamin and multimineral supplementation or placebo for one year.
Main outcome measures Primary outcomes were contacts with primary care for infections, self reported days of infection, and quality of life. Secondary outcomes included antibiotic prescriptions, hospital admissions, adverse events, and compliance.
Results Supplementation did not significantly affect contacts with primary care and days of infection per person (incidence rate ratio 0.96, 95% confidence interval 0.78 to 1.19 and 1.07, 0.90 to 1.27). Quality of life was not affected by supplementation. No statistically significant findings were found for secondary outcomes or subgroups.
Conclusion Routine multivitamin and multimineral supplementation of older people living at home does not affect self reported infection related morbidity.
Trial registration ISRCTN: 66376460.
Respiratory tract and urinary tract infections are common reasons for older people to visit their doctor,1 and viral respiratory tract infections may precipitate hospital admission in people with diabetes or cardiac or chronic obstructive pulmonary diseases.2 Ageing is associated with dysregulation of the immune system.3 Declining immunity can lead to infection and depletion of nutritional reserves, which may already be suboptimal for host defences.4
The UK national diet and nutrition survey of older people found evidence of multiple nutritional deficiencies.5 The prevalence of deficiencies on blood testing ranged from about 10% for iron, vitamin C, red cell folate, thiamine, and vitamin D in people living in the community to about 40% for vitamin C, riboflavin, and vitamin D in those living in care. Deficiencies were greater in Scotland than in southern England and greater in those aged 75 or over.
At least 25% of older people in the United Kingdom take nutritional supplements.5 Multivitamin and multimineral supplements are widely available and most brands have similar contents. Whether such supplementation influences infections in older people, particularly those not in nursing homes, is unclear from randomised controlled trials. A recent systematic review concluded that evidence for the routine use of multivitamin and multimineral supplements in older people was weak and conflicting and that further randomised controlled trials were recommended.6 That review included trials by Chandra,7 8 whose work has been questioned, leading one editor to retract a publication from one of the trials.7 9 An update to the review has been published.10
Supplementation is a low cost option for improving nutritional status. We undertook a pragmatic, randomised, placebo controlled trial to examine whether supplementation influences self reported infections, use of health services, and quality of life in people aged 65 or over.
Participants and methods
Participants were recruited from six general practices in Grampian, Scotland between February and December 2002. All people aged 65 or over who were registered with the practices were eligible, irrespective of chronic illness, unless their doctors considered them too unwell. Exclusion criteria were use of vitamin, mineral, or fish oil supplements in the previous three months (one month in the case of water soluble vitamins) or vitamin B12 injection in the past three months. Written informed consent was collected.
Participants were recruited in their practice or at home by an experienced nurse or dietician. Detailed instructions were given about trial drugs, completion of daily diaries on infection and any contacts with primary care for infections, which were returned monthly, and the need to avoid vitamin or mineral supplements from other sources during the trial.
Participants were randomised to one tablet daily of a multivitamin and multimineral supplement or a matched sorbitol placebo. Supplements were 800 μg vitamin A (acetate), 60 mg vitamin C, 5 μg vitamin D3, 10 mg vitamin E (D, L-α tocopheryl acetate), 1.4 mg thiamin (mononitrate), 1.6 mg riboflavin, 18 mg niacin (nicotinamide), 6 mg pantothenic acid (calcium D-pantothenate), 2 mg pyridoxine (hydrochloride), 1 μg vitamin B12, 200 μg folic acid, 14 mg iron (fumurate), 150 μg iodine (potassium iodide), 0.75 mg copper (gluconate), 15 mg zinc (oxide), and 1 mg manganese (sulphate). The supplement provided between 50-210% of UK reference nutrient intakes.11 Tablets were purchased from a commercial supplier and given to participants for one year.
Predefined primary outcomes were number of contacts with primary care (doctor and other primary care workers, in person or by phone) for infection, number of self reported days of infection, and health related quality of life measured by the EuroQol12 and SF-12.13 The EuroQol is a generic questionnaire used to provide a single score for health related quality of life; responses to the EuroQol were converted to a score using the EuroQol UK population tariff.14 The SF-12 is a generic measure of health related quality of life, including physical functioning, social functioning, role limitations due to physical or emotional problems, mental health, vitality, pain, and perception of general health. We used prespecified procedures to code diaries for contacts with primary care for infection and days of infection.
Predefined secondary outcomes were number of antibiotic prescriptions in primary care, number of days that antibiotics were prescribed, number of hospital admissions (including those related to infection), number of days in hospital with infection, number of infection related and all outpatient visits, adverse events reported by participants, and compliance with trial drugs (from diaries submitted monthly in all participants and tablet count at six and 12 months in a random sample of 10% of participants). Hospital data were obtained from computerised patient administration systems and hospital and general practice records.
General practice records were the only source of information on contacts with primary care and antibiotic prescriptions in over 94% of cases. Missing data were supplemented with participants' diary information when deaths subsequently occurred and the health board had removed the general practice notes, or when participants gave details of antibiotic prescriptions missing from the general practice records.
We carried out prespecified subgroup analyses for ages 65-74, 75-84, and ≥ 85 (the last two categories being subsequently combined because few participants were aged ≥ 85), sex, type of accommodation (community, or in care, such as a nursing home), and nutrition risk score (predictive of no deficiency versus one deficiency in any of four groups).
We used a nutrition assessment questionnaire for detecting older people at risk of nutritional deficiencies in vitamin C, vitamin D, folate, or iron.15 We developed questions from responses to a questionnaire and blood levels of nutritional status in a healthy local population of older people without evidence of infection.16
We determined that we would need 900 participants to provide 80% power (5% significance, two-tailed) to detect a 25% reduction in contacts with primary care for infection, where 40% of this population in the United Kingdom seek help from primary care for infections in any year.1 We assumed a dropout rate of 20%. Our trial would be able to identify, with the same power, relatively smaller effects on the other primary outcome of days of infection.
Randomisation and blinding
At recruitment, participants were randomly allocated to intervention or placebo by batch numbers generated by a password protected computer program. The program could be accessed by only the data programmer, blind to treatment allocation. The identity of the tablets was concealed in a double envelope sent by the manufacturer, which was kept locked in a cabinet during the trial. Randomisation was stratified by general practice and minimised by age (65-74, 75-84, ≥ 85), sex, and type of accommodation (community or in care).
Assessors blind to trial allocation collected participant reported primary care and hospital data. Statisticians undertook analysis also blind to trial allocation. Blinding of researchers, participants, and health professionals was maintained until completion of the analyses.
The trial was conducted according to the Medical Research Council's guidelines for good clinical practice in clinical trials.17 All data collected were subjected to quality assurance procedures. Two researchers independently entered a 10% random sample of diaries and questionnaires using a secure double data entry system. Agreement of 99% was achieved. We checked hospital records for all routinely collected data for everyone reporting a hospital admission or who had this documented in their general practice records. We entered a random sample of 10% of participants into the hospital's patient administration system to check whether any admissions had been missed; none were found. We used prespecified standardised procedures for extracting data from general practice and hospital records.
Participants were asked at the end of the study whether they knew their allocated treatment. Over 90% did not know or were unable to guess.
Each comparison was by intention to treat, irrespective of compliance with pill taking. Statistical significance for all primary and secondary end points was based on two sided tests, with 2P ≤ 0.05. We applied stricter levels of significance (2P ≤ 0.01) to subgroup analyses. For skewed continuous data, we calculated median and interquartile ranges. To account for skewness, we used negative binomial and poisson regression models to estimate the incidence rate ratio and 95% confidence intervals between groups, adjusting for the minimisation covariates, using STATA.18 We accounted for an excess number of zeroes in an outcome by using a zero inflated negative binomial or poisson model when appropriate.19 A linear regression model in SPSS was used to estimate the mean difference for health related quality of life measures between groups, after adjusting for baseline values and the minimisation covariates.20
Overall, 910 of 10 090 people (9%) aged 65 or over were recruited (fig 1). Only 13% (n = 121) of participants were lost to follow-up or reported stopping taking tablets. At least one diary was provided by 99% (901) of participants, six diaries by 93% (846), and 12 diaries by 89% (808).
The groups were well balanced at entry to the trial—for example, median age 72 and 71 in intervention and placebo groups (table 1). In total, 4% of participants were aged 85 or older; 3% lived in nursing homes. Over half of the participants were men. More than half took three or more different drugs daily. Over 93% had had an injection in the previous year to prevent influenza.
On the basis of the questionnaire on nutrition, 32% of participants in the intervention group (n = 145) and 26% in the placebo group (n = 117) were at high risk of nutritional deficiency for iron, folate, vitamin C, or vitamin D. Slightly more people in the intervention group than in the placebo group were at high risk of iron deficiency (16% versus 8%).
We found no differences between the groups for compliance with drug taking (figs 2 and 3). Compliance in participants still taking tablets and returning information in diaries was over 91% throughout the trial (fig 2). Compliance for all participants who had not died was over 78% throughout the trial (fig 3). Compliance as assessed by self report was consistent with tablet counting.
We found no statistically significant effect from supplementation on our primary outcomes of contacts with primary care staff (mostly general practitioners and practice nurses) and days of infection per person: incidence rate ratio 0.96, 95% confidence interval 0.78 to 1.19 and 1.07, 0.90 to 1.27; table 2). Supplementation did not seem to affect quality of life (table 3).
None of the secondary outcomes significantly differed between the groups (table 4).
Ancillary analyses and purported adverse events
Prespecified subgroup analyses for age, sex, and type of accommodation did not significantly differ between the groups (data not shown). Neither did we find important differences for subgroups on the basis of the nutrition risk score (no deficiency versus one deficiency in any of the four groups; table 5).
Purported adverse events were not serious (headache, insomnia, gout, for example). These events did not significantly differ between the groups (table 4). Similar numbers of participants in the groups died during the study (table 4).
Regular use of commonly available multivitamin and multimineral supplements by community dwelling older people who do not already take supplements is unlikely to reduce the number of self reported infections or associated use of health services. It remains to be seen whether populations at higher risk of infections, such as older people living in nursing homes, would benefit from supplementation.
Our population had few people aged 85 or older or in nursing homes, who may be at higher risk of nutritional deficiency.5 Our population may have been healthier than people who did not volunteer for the study. We did not, however, recruit people already taking nutritional supplements (likely to be more than 25% of this age group5). By leaving out such people, we probably excluded those with healthier diets.5 By using a simple questionnaire, we found that over 25% of the population was judged to be at high risk of nutritional deficiency and potentially more likely to benefit from supplementation, although this was not supported by our subgroup analyses. The proportion of the trial population at high risk of nutritional deficiency was similar to that in a community survey of older people in the same region of Scotland.16
Our participants were similar in age and place of residence to those studied in a previous, disputed trial that reported beneficial effects from supplementation.7 9 10 Participants in that trial had levels of nutrient deficiencies (assessed by blood sampling) similar to people designated by our questionnaire as being at high risk of nutrient deficiency. Unlike our trial, the supplement used in the previous trial had four times the levels of vitamin E and vitamin B12 and also contained selenium, magnesium, calcium, and β carotene.7
Other randomised controlled trials have not found beneficial effects of multivitamin and multimineral supplements on risk of infection in community dwelling older people.21–23 A trial in France reported no statistically significant effect of vitamin and mineral supplements on infections.21 A study in the Netherlands found that multivitamin and multimineral supplements had no significant effect on acute respiratory tract infections.22 A US study in primary care also found no significant effect from multivitamin and multimineral supplements on infections in people aged 65 and over.23
Supplementation, especially of trace elements, has been shown to reduce infections in older people in nursing care with higher levels of nutritional deficiency.24 An effect of borderline significance was reported by another trial.25 The effects of supplementation with vitamin E also seem to differ between community and nursing populations. In one study, vitamin E seemed to worsen the severity of infections in people living in the community,22 whereas it reduced the number of respiratory tract infections in a trial of nursing home residents.26
We cannot exclude the possibility that the intakes provided in the supplement were inadequate to affect the immune system. Our trial was pragmatic, which examined participants' self reported infections and use of health services rather than confirming infection by rigorous criteria seldom used inpractice. Errors in self reporting should be balanced between the two groups.
What is already known on this topic
The vitamin and mineral status of older people may be suboptimal, influencing immunity and risk of infection
It is unclear from previous trials in community dwelling older people whether multivitamin and multimineral supplements influence infection
What this study adds
Routine multivitamin and multimineral supplementation of older people living at home is unlikely to affect self reported infection related morbidity
Studies are needed in higher risk populations, such as older people living in nursing homes
Other members of the MAVIS Trial Group were Kathryn Brownie (research assistant), Janice Cruden (trial secretary), Gladys McPherson (senior information technology manager), Clare Robertson (research assistant), and Joanne Warner (research assistant). Collaborators in primary care were: Ellon Health Centre—Alan Donaldson, Ian MacKay, Duncan McKerchar, Ian Simpson, Martin Pucci, Rosamund Bell, Peter Brown, Pilar Murphy, Huber Kam, Anne Pearson, and Caroline Cumming; Gilbert Road Medical Group—John Corse, Douglas Orr, Linda Sandilands, James Scott, Murdoch Shirreffs, Sheena Tuttle, Jane White, Gordon Wilson, Jane Harvey, and Hilary Andrew; Inverurie Health Centre—James Beattie, James Black, Victor Johnston, David Hood, Jacqueline MacDonald, Sally Harkness, Fiona McKay, David Rutledge, Fiona Baxter, Gillian Brewis, Richard Gordon, Eunice Connon, Wilma Hadden; Macduff Health Centre—Iain Brooker, Alison Barbour, Pat Hoddinott, Murial Barclay, and Joy Thom; Peterhead Health Centre—Lewis Ritchie, Kenneth Strachan, Patricia Donaldson, Joyce Robertson, John Stout, Ian Small, Gregor Bruce, David Kennedy, Bruce Strachan, Graham Strachan, Dale Fenwick, Michelle Bibby, Ethel Wilson, Fiona Begg; and Queen's Road Medical Group—Iain Duthie, Geoff Clarke, Fiona Garton, Eunice Connon, Paul Davidson, Iain Stirling, Theresa Suttle, Stuart Watson, Belinda Porter, Shona Nairn, Loraine Horsburgh, Rosie Jamieson. Group members of the Data Monitoring and Safety Committee were Adam Coldwells (chair), Barbara Golden, and Lewis Ritchie. We thank the participants and staff of the general practices and medical records at Aberdeen Royal Infirmary.
Contributors AA, MKC, JAC, PCH, MMK, GMcN, ACM, CRR, DGS, AIS, and LDV (writing group of the MAVIS trial) designed the study, interpreted the data, wrote the paper, and approved the final version. AA is guarantor.
Funding Health Foundation (formerly PPP Healthcare Medical Trust). The Health Services Research Unit and Health Economics Research Unit are funded by the Chief Scientist Office of the Scottish Executive Health Department. The views expressed are those of the authors.
Competing interests None declared.
Ethical approval Grampian research ethics committee.