Reference
Mursu J, Robien K, Harnack LJ, et al. Dietary supplements and mortality rate in older women. Arch Intern Med. 2011;171(18):1625-1633.
Design
Observational study assessing the relationship between supplement use and total mortality rate in older women. The study utilized data from the Iowa Women’s Health Study, which was originally designed to examine associations between several host, dietary, and lifestyle factors and the incidence of cancer in postmenopausal women.
Participants
38,772 older Caucasian women with a mean age of 61.6 years at baseline
Study Medication and Dosage
Supplement use was self-reported in 1986, 1997, and 2004. Participants filled out self-administered questionnaires assessing the use of 15 supplements, including a multivitamin and individual nutrients. Dose-response associations were computed for selected supplements.
Outcome Measures
Mortality rate identified through the State Health Registry of Iowa and the National Death Index
Key Findings
The authors concluded that several commonly used dietary vitamin and mineral supplements may be associated with increased total mortality risk; associations were strongest with supplemental iron. Calcium use was significantly associated with a significantlydecreased risk of mortality.
Compared with non-users, supplement users had a lower BMI and waist-to-hip ratio and lower prevalence of diabetes mellitus, high blood pressure, and smoking status. Supplement users had a higher education level, were more physically active, and were more likely to use estrogen replacement therapy. Furthermore, supplement users were more likely to have lower intake of energy, total fat, monounsaturated fatty acids, and saturated fatty acids and were more likely to have higher intake of protein, carbohydrates, polyunsaturated fatty acids, alcohol, whole-grain products, fruits, and vegetables. These data demonstrate that supplement users engage in healthy habits and have better biomarkers of health status when compared to non-users.
The investigators applied a variety of statistical adjustments to the data in an attempt to separate the effect of healthy habits and biomarkers of health status from the effects of the nutrients on total mortality. A minimally adjusted statistical analysis (adjusted for age and energy intake) demonstrated significant associations between lower risk of total mortality and intakes of vitamin B complex; vitamins C, D, and E; and calcium compared with nonuse of these nutrients. However, calcium use alone remained significantly associated with lowered risk of mortality after further statistical adjustments were applied to account for education level, place of residence, diabetes mellitus, high blood pressure, body mass index, waist-to-hip ratio, hormone replacement therapy, physical activity, and smoking status.
Minimally adjusted analysis also demonstrated associations between magnesium and zinc and lowered risk for mortality; however, once other biomarkers of good health and healthy lifestyle factors were adjusted for, these nutrients became associated with increased risk. Multivitamins, vitamin B6, copper, and iron were associated with an increased risk of mortality throughout all statistical analysis.
Implications for Clinical Practice
The main impact of this study on clinical practice will likely result from increased questions and concerns from confused patients. Media coverage has sensationalized findings that suggest increased total mortality risk without discussing the study’s limitations or placing it in context with other evidence that supports the benefits of nutrient supplementation1–4 or larger studies with a similar methodology that show no effect on total mortality risk.5
It is important to keep in mind that this is an observational study that demonstrates correlations—not cause and effect. Observational studies include some limitations that make sweeping generalizations inappropriate. For example, participants in this study were asked about dietary supplement intake, but the accuracy of answers provided in questionnaires was not verified and participants were never asked why they were taking supplements. Individuals are known to use or start using dietary supplements because of illness—the so-called sick-user effect—which influences mortality rates.4 These variables are difficult to account for in study design.
In this study, data derived from questionnaires are statistically manipulated in an attempt to isolate effects of vitamin and mineral supplementation from effects of other factors that influence total mortality, such as healthy diet, healthy body weight, and exercise. This reductionist model attempts to statistically tease out one piece of the healthy equation, which complicates interpreting the relevance of results. For example, when the authors did the initial minimally adjusted analysis, it appears there is benefit for many of the nutrients studied, not just calcium; yet instead of stopping there, they went on to “further adjust” the data, eliminating the signs of benefit.
In addition to the aforementioned limitations of observational studies, this study is flawed at its foundation. The classification of participants as “users” and “non-users” of particular dietary supplements is not matched to the types of nutrients to which participants were exposed. This results in an incorrect set of comparisons.
The “non-user” category includes only “non-users” of a particular supplement and is not truly a set of “non-users of supplements.” For example, the category of vitamin B6 “non-users” included participants who did not take an isolated vitamin B6 supplement, but it did not exclude participants who used a multivitamin or other dietary supplement that contained vitamin B6. This is noteworthy for several reasons. First, there were 1,269 “users” of vitamin B6 alone; however, the study’s 12,769 “users” of multivitamins were counted as “nonusers” of vitamin B6. This means at least 11,500 participants were considered “non-users” of vitamin B6 when, in fact, they were exposed to vitamin B6 in a multivitamin. Therefore, the findings are not related to vitamin B6 exposure and cannot be used to support the authors’ conclusion that taking vitamin B6 is associated with an increased risk of total mortality.
Second, when the data are analyzed with these criteria, the number of “users” of a particular supplement is very small compared to “non-users” of that particular supplement. This is important because a small number of deaths in a particular “user” category can statistically demonstrate increased risk of mortality when compared to “nonusers.” For example, increased risk of mortality was demonstrated for copper with only 229 “users” as compared to 38,772 “non users.” Despite the fact that it is perplexing why these post menopausal women would be supplementing copper in isolation, comparing total mortality for this very small group of 229 participants to the much larger group of 38,772 is nonsensical.
It is also noteworthy that the authors never report total mortality for the entire category of vitamin and mineral “users” (n=24,329) vs. “non- users” (n=14,443), which is balanced mathematically and is a relevant association to report.
Research consistently shows that dietary supplement users are more likely than non-supplement users to engage in other healthy practices.
The flawed methodology used to identify “users” and “non-users” (Table 2 in the original paper) is used throughout the data analysis. Given the fact that the cohort of supplement users demonstrated improved biomarkers of good health and healthy lifestyle habits, an overall comparison of “users” to “non-users” of dietary supplements is an important omission in the manuscript. In fact, if such a comparison demonstrated decreased or no change in risk of total mortality, this finding would significantly weaken the relevance of individual findings for multivitamins, vitamin B6, copper, and iron. A finding of reduced or decreased risk for all “users” combined would weaken the strength of the authors’ sweeping conclusions about risks associated with vitamin and mineral use.
The manuscript excluded a discussion of recent studies on the same topic. For example, the “Multivitamin use and the risk of mortality and cancer incidence: the multiethnic cohort study” by Park et al was a similar, but much larger, observational study published in 2011 that assessed the relationship between multivitamin use and mortality and cancer.5 The study included 182,099 participants, both men and women, from multiple ethnic backgrounds (in comparison, the study by Mursu et al, included 38,772 older Caucasian women). Park et al found no associations between multivitamin use and mortality from all causes, cardiovascular diseases, or cancer in “users” vs. “non-users.” The findings did not vary across subgroups by ethnicity, age, body mass index, pre-existing illness, single vitamin/mineral supplement use, hormone replacement therapy use, or smoking status. There also was no evidence indicating that multivitamin use was associated with risk of cancer, overall or at major sites such as lung, colorectum, prostate, and breast. In conclusion, there was no clear decrease or increase in mortality from all causes, cardiovascular disease, or cancer, nor in morbidity from overall or major cancers among multivitamin supplement users.
Integrative clinicians should be reassured that research consistently shows that dietary supplement users are more likely than non-supplement users to engage in other healthy practices. This study did not discount those facts, and expressly noted that supplement users were more likely to be physically active and have a lower BMI and waist-to-hip ratio, and were less likely to smoke or have high blood pressure or diabetes mellitus.
The inherent limitations of observational studies, confounded by the nonsensical classification of “users” and “non users” by Mursu et al makes the findings of the current study questionable at best. The manuscript lacks balance, as evidenced by no discussions of similar recent studies, and includes statements of the authors’ opinions, such as, “We recommend that [dietary supplements] be used with strong medically based cause, such as symptomatic nutrient deficiency….” This basically means these researchers would rather wait until a scurvy diagnosis before acknowledging any need for supplemental nutrients. These factors suggest a potential author bias. While we all have our biases, it is irresponsible to wait until patients suffer “symptomatic nutrient deficiency” before taking action. In conclusion, the findings reported by Mursu et al should be interpreted with caution and do not justify changes to integrative clinicians’ recommendations regarding multivitamins or other nutrients.
