Usage patterns, health, and nutritional status of long-term multiple dietary supplement users: a cross-sectional study

Background Dietary supplement use in the United States is prevalent and represents an important source of nutrition. However, little is known about individuals who routinely consume multiple dietary supplements. This study describes the dietary supplement usage patterns, health, and nutritional status of long-term multiple dietary supplement users, and where possible makes comparisons to non-users and multivitamin/mineral supplement users. Methods Using a cross-sectional study design, information was obtained by online questionnaires and physical examination (fasting blood, blood pressure, body weight) from a convenience sample of long-term users of multiple dietary supplements manufactured by Shaklee Corporation (Multiple Supp users, n = 278). Data for non-users (No Supp users, n = 602) and multivitamin/mineral supplement users (Single Supp users, n = 176) were obtained from the National Health and Nutrition Examination Survey (NHANES) 2001–2002 and NHANES III 1988–1994. Logistic regression methods were used to estimate odds ratios with 95% confidence intervals. Results Dietary supplements consumed on a daily basis by more than 50% of Multiple Supp users included a multivitamin/mineral, B-complex, vitamin C, carotenoids, vitamin E, calcium with vitamin D, omega-3 fatty acids, flavonoids, lecithin, alfalfa, coenzyme Q10 with resveratrol, glucosamine, and a herbal immune supplement. The majority of women also consumed gamma linolenic acid and a probiotic supplement, whereas men also consumed zinc, garlic, saw palmetto, and a soy protein supplement. Serum nutrient concentrations generally increased with increasing dietary supplement use. After adjustment for age, gender, income, education and body mass index, greater degree of supplement use was associated with more favorable concentrations of serum homocysteine, C-reactive protein, high-density lipoprotein cholesterol, and triglycerides, as well as lower risk of prevalent elevated blood pressure and diabetes. Conclusion This group of long-term multiple dietary supplement users consumed a broad array of vitamin/mineral, herbal, and condition-specific dietary supplements on a daily basis. They were more likely to have optimal concentrations of chronic disease-related biomarkers, and less likely to have suboptimal blood nutrient concentrations, elevated blood pressure, and diabetes compared to non-users and multivitamin/mineral users. These findings should be confirmed by studying the dietary supplement usage patterns, health, and nutritional status of other groups of heavy users of dietary supplements.


Background
Diet and nutrition play important roles in the maintenance of health and prevention of disease [1,2]. Dietary supplements represent an important source of essential nutrients since they are widely used and often contain 100% or more of the Daily Value of one or more nutrients [3,4]. They are also of concern because of potential adverse effects [5]. Prior studies have reported the prevalence of dietary supplement usage and the characteristics of users in the United States (US) population. In the most recent reporting of nationwide survey data, the National Health and Nutrition Examination Survey (NHANES) 1999-2000, 52% of adults reported taking a dietary supplement in the past 30 days [4]. This and other surveys have generally found that dietary supplement usage is more common in women than men, in older participants than in younger ones, in Whites compared with Blacks and Mexican Americans, in the more educated as compared to the less educated, and in the more affluent as compared to less affluent [4]. In NHANES 1999-2000, approximately 47% of dietary supplement users reported taking only one type of supplement (most commonly a multivitamin/mineral) [4]. Only 3 individuals out of over 11,000 surveyed reported taking 20 or more different supplements in the past 30 days. Consequently, little if any descriptive information has been reported about individuals who typically consume multiple dietary supplements.
This cross-sectional study was undertaken to describe the dietary supplement usage patterns, serum nutrient and biomarker concentrations, and health status of a convenience sample of individuals who were daily users of multiple dietary supplements (median of 26 different dietary supplements taken daily in the prior 12 months). In addition, biomarker concentrations and the health status of multiple dietary supplement users were compared with two other convenience samples assembled from NHANES: non-users of supplements and those who consumed a multivitamin/mineral supplement only.

Participants
To assemble a sample of long-term users of multiple dietary supplements, individuals who had been consumers of dietary supplements for ≥ 20 years from a dietary supplement manufacturer and distributor (Shaklee Corporation, Pleasanton, CA) were invited to participate in the study. The invitation was sent by electronic mail to approximately 1,200 individuals meeting this criterion. Of the total invitees, 435 successfully completed online questionnaires to obtain information about height, annual household income, education, medical history and current medical conditions, and dietary supplement usage. Reasons for non-participation were not investi-gated. A subset of 300 individuals who completed the questionnaires and were free of cancer, other than nonmelanoma skin cancer, were randomly chosen and asked to attend a physical examination for the purpose of providing a 12-hour fasting blood sample and having their body weight and blood pressure measured. A final analytic sample of 278 participants had complete questionnaire and examination data. These individuals composed the 'Multiple Supp' users. The study was reviewed and approved by an independent institutional review board (Health Research and Studies Center, Palo Alto, California), and all participants provided informed consent.

Data collection
For the Multiple Supp users, online questionnaires were used to obtain information about height, annual household income, education, medical history and current medical conditions, and current dietary supplement usage. Questions about medical history and current medical conditions were patterned after questions asked in NHANES 2001NHANES -2002. The dietary supplement questionnaire included 57 dietary supplements and foods sold by Shaklee in 2005 and asked about usual intake in the previous 12 months. The questionnaire included four intake frequency categories including 'rarely or never used,' '1-3 days per week,' '4-6 days per week,' and 'every day,' and also asked about usual serving size.
Physical examinations were conducted during a national meeting of Shaklee product consumers held in Chicago, Illinois in August 2005. Data collection took place between July and August of 2005. Body weight was measured using a physician's scale with shoes and heavy clothing removed. Triplicate blood pressure measurements were taken by an automated Welch Allyn device (Model 5201) with an appropriate-sized cuff after subjects were seated quietly in a chair for 2 minutes, with feet on the ground and with the arm at the level of the heart. Twelvehour fasting blood samples were collected, processed, aliquoted and stored frozen at -80°C. Samples were assayed for red blood cell (RBC) folate, serum ascorbic acid stabilized using 10% metaphosphoric acid, alpha tocopherol, carotenoids, retinol, 25-hydroxyvitamin D, ferritin, C-reactive protein (CRP), lipids, and homocysteine (Table 1). Assay procedures followed those that were utilized in NHANES 2001-2002 and in NHANES III [6-12] as described in Table 1.

Multivitamin/mineral and non-supplement users from NHANES
We used NHANES (NHANES III and 2001-2001) data as a source of comparison data for multivitamin users and non-users of supplements. NHANES is designed to monitor the health and nutrition status of the US population and participation consisted of an in-home interview and an examination in the NHANES mobile examination unit [6]. Dietary supplement intake was assessed in NHANES by asking participants about their dietary supplement usage patterns in the past 30 days prior to their home interview, including use of any vitamins, minerals, or other dietary supplements. NHANES oversamples certain groups, such as older people and low-income individuals, in order to obtain greater precision in estimates for those groups; sample weights are applied by NHANES to adjust for oversampling and nonresponse. NHANES sample selection involves a stratified multistage probability design involving counties, blocks, and households. Consequently, variance estimates require use of special software making use of NHANES strata and primary sampling units to account for the sampling design.
To match the age and race composition of the Multiple Supp users, we selected from NHANES 2001-2002 participants all of the White men and women, >35 years of age, and free of cancer other than non-melanoma skin cancer, who met certain supplement use criteria. Only non-supplement users and users of supplements containing vitamins and/or minerals were included for the present analysis. We identified 602 NHANES participants who consumed no dietary supplements in the 30 days prior to the home interview; these became the 'No Supp' users. We also identified 176 individuals who consumed a multivitamin/mineral and no other dietary supplements, and did so at least 15 days over the past 30 days. These became the 'Single Supp' users. To obtain the most recent data possible for comparative purposes, RBC folate, and serum ferritin, homocysteine, CRP, and lipid data were obtained from NHANES 2001NHANES -2002. However, serum retinol, ascorbic acid, alpha tocopherol, and serum carotenoid data were not available from NHANES 2001NHANES -2002NHANES or 2003NHANES -2004. Therefore, data for these variables for Single Supp and No Supp users who were White, >35 years of age, and free of cancer other than non-melanoma skin cancer were obtained from NHANES III, 1988-1994. The geographical distribution of the individuals in the three users groups is not known. In addition, the duration of dietary supplement usage by NHANES participants, aside from the 30 days prior to their home interview, is unknown.

Statistical analyses
After combining the Multiple Supp users with the Single Supp and No Supp users from NHANES, data were reweighted. The weights for the NHANES groups were calculated by dividing each NHANES weight by the sum of the NHANES weights of the user group. Appropriate weights were used depending on whether the data came from the interview or the Mobile Examination Center. Multiple Supp users were assigned a weight of 1. Strata and primary sampling units from NHANES were used for those user groups. For the Multiple Supp users, a new stratum variable was assigned, and each member of the Multiple Supp group was assigned to a unique primary sampling unit. These adjustments permit more accurate variance estimates and account for stratification factors. However, the Single Supp and No Supp users from NHANES should not be interpreted as being nationally representative samples. Calculations used SUDAAN Ver-  Differences in the characteristics of the supplement user groups were evaluated using chi-square methods for categorical variables and one-way analysis of variance for continuous variables. Multiple regression techniques were employed to examine differences in nutrient and biomarker concentrations of user groups. Comparisons of mean nutrient concentrations among user groups (RBC folate, and serum retinol, ascorbic acid, alpha tocopherol, carotenoids, and ferritin) were adjusted for sex and age. Comparisons of biomarker concentrations among user groups (serum homocysteine, CRP, total cholesterol, and lipids) were further adjusted for age 2 , education category (<high school, high school graduate, or >high school), income category (<$30,000, ≥$30,000 and <$40,000, ≥$40,000 and <$50,000, ≥$50,000 and <$60,000, ≥$60,000 and <$70,000, and ≥$70,000) and income 2 , and body mass index (BMI). For outcomes treated as dichotomous variables, logistic regression methods were used to estimate odds ratios (OR) with 95% confidence intervals (CI), adjusted for sex, age and age 2 , education category, income category and income 2 , and BMI. The model fit was examined using the Hosmer and Lemeshow Goodness-of-Fit test. The referent group for risk estimation was the No Supp user group. Statistical significance was defined as p < 0.05.

Results
Significant differences by user group were found for sex, age, BMI, education, and household income (   (Table 3). In addition, a majority of women also consumed gamma linolenic acid and a probiotic supplement, whereas men also consumed zinc, garlic, a soy protein supplement, and saw palmetto daily.

Serum nutrient and biomarker concentrations
There was a statistically significant increase in RBC folate and serum concentrations of ascorbic acid, alpha and beta carotene, and alpha tocopherol with degree of supplementation use ( There was a decrease in concentrations of biomarkers associated with disease risk as supplement use increased (Table 4). Across the three user groups there was a decrease in serum CRP and the ratio of total cholesterol to HDL-cholesterol, and an increase in HDL-cholesterol concentration. Serum homocysteine and triglycerides were lower in the Multiple Supp users than in the other two groups. Finally, serum total cholesterol was significantly lower in the Multiple Supp group, whereas LDL-cholesterol concentrations did not vary significantly by supplement use.

Blood pressure and adverse biomarker concentrations
Adjusted risk estimates for elevated blood pressure and suboptimal or elevated concentrations of serum biomarkers are reported in Table 5

Health status and disease prevalence
Prevalence of self-reported disease was low in each user group and most risk estimates did not reach statistical significance (Table 6). An exception was the reduced risk of diabetes in Multiple Supp users (OR = 0.27; 95% CI: 0.12-0.59). Risk of coronary heart disease was also lower in this group (OR = 0.48; 95% CI: 0.22-1.09), but was only marginally statistically significant. For self-assessed health status, Multiple Supp users, but not Single Supp users, were less likely to report their health as 'good, fair, or poor' versus 'excellent or very good' (OR = 0.26; 95% CI: 0.17-0.41) compared to the No Supp users group (Table 6).

Discussion
Little has been reported about daily users of multiple dietary supplements. We assembled a sample of long-term users of multiple dietary supplements for the purpose of describing their supplement usage patterns. On average, Multiple Supp users in this study consumed 17 different supplements every day. Dietary supplements consumed on a daily basis by more than 50% of Multiple Supp users included a multivitamin/mineral, B-complex, vitamin C, carotenoids, vitamin E, calcium with vitamin D, omega-3 fatty acids, flavonoids, lecithin, alfalfa, coenzyme Q10 with resveratrol, glucosamine, and a herbal immune supplement. Women also consumed gamma linolenic acid and a probiotic supplement, whereas men also consumed zinc, garlic, a soy protein supplement, and saw palmetto ( Table 2). By comparison, in NHANES 1999-2000, 52%  This fact has important implications for nutrition intervention trials, some of which have permitted participants, including those in the control group, to take a multivitamin/mineral supplement [25]. Our findings suggest that this practice may weaken the power of trials to detect treatment effects of the study supplement.

4: Serum nutrient and biomarker concentrations of a sample of long-term users of multiple dietary supplements (Multiple Supp), and of multivitaminmineral users (Single Supp) and non-users (No Supp) from NHANES 2001-2002 and NHANES
Degree of supplement use was associated with several markers of disease risk. Elevated serum homocysteine concentrations were found in approximately 45% of No Supp users, 37% of Single Supp users, and 11% of Multiple Supp users, despite the fact that the data were obtained after the 1998 fortification of grain products with folate, * Logistic regression methods were used to estimate odds ratios (OR) with 95% confidence intervals (CI), adjusted as in footnote to Table 4. The referent group for risk estimation was the No Supp group. † OR and 95% CI not estimable because zero events in the user group. Number and percent shown instead. HDL = high-density lipoprotein, LDL = low-density lipoprotein, DBP = diastolic blood pressure, SBP = systolic blood pressure. and despite the low prevalence of inadequate RBC folate in any of the user groups. If moderately-elevated serum homocysteine is shown to be an important cardiovascular disease risk factor, the findings of this study suggest that intake of the relevant B vitamins above what is typically found in a multivitamin/mineral supplement, and/or the diet, may produce a more favorable homocysteine-lowering effect.
Circulating CRP concentration has been shown to be predictive of future cardiovascular disease risk in prospective studies among asymptomatic individuals and may have a direct effect on the progression of atherosclerosis [26]. In the present study, serum CRP concentrations decreased and serum ascorbic acid increased with increasing degree of supplement use (Tables 2 and 4). This finding is consistent with recent reports that vitamin C supplementation lowers plasma CRP concentrations [27], and that dietary intakes of vitamin C and plasma ascorbic acid concentration are inversely associated with plasma CRP [28].
Studies are underway to confirm the impact of vitamin C supplementation on circulating CRP and other markers of immune function.
Serum ascorbic acid concentration >1.0 mg/dL has been suggested as optimal relative to reduced risk of cardiovascular disease and cancer [24]. Supplement use was also associated with lower serum triglycerides and higher HDL-cholesterol concentrations. In consequence, the risk of an elevated ratio of total cholesterol to HDL-cholesterol was significantly lower in the Multiple Supp group compared to the No Supp group (Table 5). Prevalent use of fish oil supplements in the Multiple Supp users may explain these findings as omega-3 fatty acids in fish oil have been shown to increase HDLcholesterol and decrease triglyceride concentrations [29].
Although beyond the scope of the present study, lower risk of cardiac arrhythmias may be another benefit of consumption of omega-3 fatty acids [30,31].
Self-assessed health status has been found to be a remarkably good marker of prospective health outcomes [32]. Compared to No Supp users, Multiple Supp users were more likely to describe their health as 'very good' or 'excel-  (Table 5). Dietary interventions have been shown to reduce blood pressure [33]. However, clinical trials of individual nutrients have typically found small and inconsistent effects [33]. Dietary folate and vitamin C, as well as plasma ascorbic acid, have been found to be inversely associated with blood pressure in observational studies [34,35]. However, intervention trials with vitamin C have yielded inconsistent results [35], while in two small trials, folic acid has been effective at lowering blood pressure [36,37]. The finding of a lower risk of elevated blood pressure in the Multiple Supp group, which also had the highest concentrations of RBC folate and serum ascorbic acid, suggests that the relationship between these nutrients and blood pressure may warrant further investigation.
The lower risk of diabetes in the Multiple Supp group is consistent with evidence that oxidative stress may be a mechanism linking insulin resistance with dysfunction of pancreatic beta cells and endothelial dysfunction, eventually leading to diabetes [38]. Although biomarkers of oxidative stress were not available in the three user groups, serum concentrations of antioxidants including ascorbic acid, carotenoids, and alpha tocopherol were all significantly higher in the Multiple Supp users compared to the other groups (Table 2).
An important limitation of the study is the fact that the data are cross-sectional, and therefore the reported associations, particularly with health outcomes (i.e., blood pressure and diabetes), cannot presume causality. Also, the three user groups should not be interpreted to represent unbiased national estimates. In addition, although we adjusted for potential confounders such as age, sex, income, education, and BMI, residual confounding could possibly account for the findings. Better access to health care and variables related to higher socioeconomic status in the supplement user groups (e.g., healthier diets and lifestyles) are logical hypotheses that may account for these results. However, the No Supp and Single Supp user groups were similar with respect to education and income (Table 2), so these factors may not explain the significantly better nutrient and C-reactive protein concentrations seen for Single Supp users in Table 4, and homocysteine and HDL risk levels (Table 5). Furthermore, the lack of differences across the three user groups in biomarkers such as serum total cholesterol and LDLcholesterol suggest that factors related to health care access and socioeconomic status do not fully explain the results.
This study is the first we are aware of to describe the usage patterns of long-term users of multiple dietary supplements, an unusual sample that cannot be captured in national representative surveys. While only 3 individuals out of over 11,000 surveyed in NHANES 2001-2002 had taken 20 or more kinds of dietary supplements in the past 30 days, in our Multiple Supp users, 87% of the sample reported having taken 20 or more kinds of supplements daily. Thus, this sample of Multiple Supp users affords a rare opportunity to understand a segment of the population that take multiple dietary supplements, and to understand the nutrient and other correlates of this practice. In addition, Multiple Supp users were long-term consumers of dietary supplements. Thus, their serum nutrient and biomarker concentrations were likely to be representative of their long-term supplementation patterns.

Conclusion
This sample of long-term users of multiple dietary supplements was found to consume a broad array of vitamin/ mineral, herbal, and condition-specific dietary supplements on a daily basis. They were more likely to have optimal concentrations of biomarkers associated with reduced disease risk, and less likely to have suboptimal circulating nutrient concentrations, elevated blood pressure, and diabetes than multivitamin/mineral users and non-users. There was no evidence that intake of vitamin D in multiple supplement users was excessive based on serum 25-hydroxyvitamin D concentrations. These findings should be confirmed by studying the dietary supplement usage patterns and health and nutrition status of other groups of heavy users of dietary supplements. Our study findings should also be weighed in the context of recent randomized controlled trials and related metaanalyses [39,40] which have raised concern about potential detrimental effects of select dietary supplements, particularly beta carotene and alpha tocopherol. While the typical daily intake of beta carotene from supplements in the Multiple Supp group was low relative to those studies (approximately 5000 IU/d of vitamin A as beta carotene), the typical intake of alpha tocopherol was 800 IU/d, similar to the dose in a number of randomized trials. Further research on usage patterns of dietary supplements and the potential health effects of dietary supplements is needed. The present report serves as a baseline to which the findings of subsequent monitoring efforts can be compared.
Abbreviations US = United States, NHANES = National Health and Nutrition Examination Survey, RBC = red blood cell, CRP = C-reactive protein, BMI = body mass index, OR = odds