Trends in Nutrient and non-Nutrient containing Dietary Supplement Use among U.S. Children from 1999–2016

Abstract

Objectives:

To characterize dietary supplement usage among U.S. children, including product type, motivations, user characteristics, and trends over time with a primary focus on non-vitamin/non-mineral dietary supplements (NVNM).

Study design:

Overall, NVNM, and vitamin and/or mineral dietary supplement use; motivations for use, and trends in use over time were examined in children (≤19y) using the NHANES 1999–2016 data (n=42,510).

Results:

Between 1999 and 2016, overall DS and VM use among all children remained relatively stable at ~30%; yet, NVNM use increased from 2.9% to 6.4%, mainly due to increased use of omega-3 polyunsaturated fatty acids. NVNM use was higher in boys than in girls (3.9% vs 3.3%), and higher in older children than in younger children (p_trend<0.0001) – opposite of what was observed with VM use. Although both user groups shared two primary motivations, both motivations were reported by a significantly higher percent of VM users vs. NVNM users: to maintain health (38.7% vs. 23.1%) and to improve health (33.1% vs. 22.6%). NVNM users were much more likely to use DS for relaxation, stress, and sleep; for mental health; and for colon and bowel health.

Conclusions:

Although the prevalence of any DS and VM use among U.S. children have both remained stable, the prevalence of NVNM use has increased substantially over time. Yet, NVNM use remains relatively low overall. NVNM use exhibited different patterns by sex, age, and motivations when compared with VM use. Despite increasing NVNM use, high quality evidence supporting their use is lacking, especially in children.

More than one-half of U.S. adults and around one-third of infants, children, and adolescents (henceforth children) report using at least one dietary supplement (DS), and many report taking multiple products.^1–6 The Dietary Supplement Health and Education Act of 1994, a legislation that defines and regulates DS, defines them as “any vitamin, mineral, herb or other botanical, or amino acid, or a concentrate, metabolite, constituent, extract, or a combination of these ingredients that seek to increase total intake”.⁷ Yet, it does not provide guidance on defining specific types of DS products.⁸

Previous research has characterized patterns of DS use overall – mainly cross-sectional reports on both micronutrient containing (i.e., vitamins and/or minerals, VM) and non-vitamin, non-mineral (NVNM) supplements – among U.S. adults ^{2, 5, 9} and children.^{3, 6, 10–12} However, among children, very little data has been published on DS use patterns over time. Therefore, the purpose of this analysis was to characterize types and trends of DS, especially NVNM products, commonly consumed by U.S. children along with motivations for their use using nationally representative, National Health and Nutrition Examination Survey (NHANES) data. NVNM within this analysis were inclusive of DS products that do not contain, or contain a very limited amount of VM. Patterns of DS usage were also examined by demographic, socioeconomic, and lifestyle factors.

Methods

This study presents data from the 1999–2016 NHANES, a nationally representative, continuous, cross-sectional survey of noninstitutionalized U.S. civilians, conducted by the Center for Disease Control and Prevention’s (CDC) National Center for Health Statistics (NCHS).¹³ The NHANES has a complex and stratified multistage probability sampling design, and uses analytic survey weights to account for nonresponse and oversampling of some groups. The NHANES protocol (and publicly released de-identified data) was approved by the Research Ethics Review Board and written informed consent was obtained for all participants or their proxies at the CDC/ NCHS. Nine survey cycles of data were combined to include all U.S. children aged ≤19y (n=42,469), inclusive of the years from 1999 to 2016. Children with missing data on DS use (n=42) were excluded, yielding a final analytic sample size of 42,510 children.

In NHANES, data are collected in two phases: an in-home interview where data on health information and demographics are first collected followed by a physical health examination in the Mobile Examination Center (MEC), a traveling clinic. Demographic, lifestyle, and socioeconomic characteristics are collected via the Computer-Assisted Personal Interview system during the in-home interview. For participants <16y of age or for those who are unable to provide their own information, a proxy is assigned.

The NHANES sampling age framework was applied to this analysis: 0–5y, 6–11y, and 12–19y of age. Prevalence of DS use among infants aged <1y was reported separately as these children had a distinct pattern of DS use compared with children 1–5y of age. In accordance with the NHANES protocol, race and Hispanic origin groupings were defined by NCHS as: non-Hispanic white, non-Hispanic black, Hispanic (including Mexican American), and other races, including multi-racial. Family income-to-poverty ratio (PIR) – the ratio of household income to the poverty threshold – was used as a proxy for family income, adjusted for family size and inflation. A PIR <130% is the cutoff to determine financial eligibility for the federal Supplemental Nutrition Assistance Program; therefore, consistent with previous studies, PIR was categorized as: <130%, 130–350%, and >350%.^{3, 14, 15} Health insurance was categorized as uninsured, private, or public at the time of survey collection. Those covered under both private and public insurance plans were included in the private health insurance category.

Screen time and body mass index (BMI) data was only available for participants ≥2y of age. Screen time was examined as a proxy for physical activity; it was calculated using the Physical Activity Questionnaire as the total number of hours per day spent looking at a screen (television, computer, or video game) and categorized as ≤1 hour/day, >1 to 2 hours/day, >2 to 4 hours/day, and >4 hours/day. BMI was calculated as kg/m², using measurements of height (in meters) and weight (in kilograms) collected by trained technicians in the MEC as part of a physical examination. BMI percentiles were used to categorize each participant’s weight status as underweight (<5^th percentile), healthy weight (5^th to <85^th percentile), overweight (85^th to <95^th percentile), or obese (≥95^th percentile) according to the growth charts developed by the CDC for children 2–18y of age.¹⁶ For children >18y of age, CDC defined BMI cut-offs were used to categorize each participant’s weight status as underweight (<18.5 kg/m²), normal weight (18.5 to <25kg/m²), overweight (25.0 to <30 kg/m²) or obese (≥30 kg/m²).¹⁷

Dietary Supplement Information and Categorization

In NHANES, information on DS use over the past 30 days is collected through an in-home inventory and participant questionnaire known as the Dietary Supplement Questionnaire (DSQ), during the in-home interview (phase I). Participants are asked to show product containers for all DS taken in the past 30 days. Trained interviewers record the name and manufacturer from the product labels of each DS product reported. If DS product labels are not present (<12% of the time), a verbal report of these details are recorded. Trained nutritionists at the NCHS then review this data, obtain labels for supplements reported verbally, and compile this information into a product-level database. This database, known as the NHANES Dietary Supplement Database, and the DSQ participant reports for each cycle are all available online through the NHANES website.¹⁸ Additional details for the survey protocol can also be found on this site.¹⁸

Supplement use reported on the DSQ was defined as the use of any vitamins, minerals, or other DS use over the previous 30 days. Based on methods described in previous studies, all DS were classified into two primary and mutually exclusive categories – VM containing DS and NVNM containing DS.^{2, 3, 10, 19} If a supplement contained both VM and NVNM ingredients, a careful, product-specific review was conducted, and the supplement was classified according to its primary use and/or primary ingredient (Table 4; available at www.jpeds.com). For example, a DS containing mainly micronutrients with a small amount of omega-3 fatty acids, and primarily taken as a multivitamin, was classified as a VM. On the other hand, a DS containing cranberry concentrate with added vitamin C, and primarily taken for the antioxidant properties of cranberry, was classified as an NVNM. The majority of supplements that contained VM in combination with one or more NVNM product were primarily used for their VM content. These combinations usually contained lesser amounts of the NVNM ingredient than typically consumed when the NVNM was reported as an individual product or as part of a targeted blend. Consequently, unless they were the primary ingredient in such a combination, NVNM products were exclusive of VM. Based on previous literature, NVNM products were further classified into specific subcategories, including omega-3 polyunsaturated fatty acids (PUFAs), botanicals (inclusive of herbals), probiotics, and fiber and are further outlined in Table 4.² For botanical supplements, we differentiated between DS containing single and multiple botanicals. The one exception was echinacea, which was not mutually exclusive of any other category due to low prevalence of its use. For this analysis, VM were aggregated into one category inclusive of multivitamin-mineral products and single nutrient-containing DS because the prevalence of single micronutrient supplement use is low.³ Although the inclusion of echinacea was data driven, overall VM and NVNM categories and all subcategories were based on previous NHANES studies in adults and children.^{2, 3, 5}

Table 4 online.

Classification system used to create dietary supplement categories, NHANES 1999–2016

Product Category	Definition	Examples
Vitamin-mineral/micronutrient (VM)	Any product containing one or more vitamins and/or one or more minerals as a primary ingredient (or for primary use); may contain small amounts of non-vitamin, non-mineral compounds	Sundown Kids multivitamin, Nature’s Plus high potency chewable iron with vitamin C and herbs
Non-vitamin, non-mineral (NVNM)	Any product containing botanicals/herbs, omega-3 PUFAs,a probiotics, fiber, joint, phosphatidyl choline, protein/amino acids, enzymes, melatonin, red yeast, brewer’s yeast, lipoic acid, carnitine, DHEA, CLA, bee pollen, gelatin, colostrum, creatine, cartilage, SAME, HTP, hyaluronic acid, collagen, malic acid, GABA, DMAE, betaine, caffeine, or kelp as a primary ingredient (and for primary use); may contain small amounts of vitamins and/or minerals	Puritan’s Pride turmeric, Dolphin Pals DHA gummies for kids, Spring Valley probiotic acidophilus

^aOmega-3 PUFAs, polyunsaturated fatty acids; includes omega-3–6-9 PUFA containing supplements in which omega-3 was in higher quantity than the others. Also includes cod liver oil supplements that did not list “omega” or “fat” in the Dietary Supplement Database: Ingredient Information. These inclusions increased omega-3 PUFA DS by n=14 across all 18 years.

Additional analyses were completed to examine motivations for use of NVNM among children; however, these data were only available in NHANES 2007–2016 (n=785). Although the data available were generally consistent across these years, differential classification was used for some motivations, and is denoted in the tables.

Statistical Analyses

Data were analyzed using SAS (version 9.4; SAS Institute, Inc, Cary, NC, USA) and SAS-callable SUDAAN (version 11; RTI International, Research Triangle Park, NC, USA) software programs to properly account for survey weights and complex survey design, non-response, non-coverage, and planned oversampling of some population subgroups. All NHANES cycles covering 18 years (inclusive of 1999–2016) were used to describe prevalence of DS use, including NVNM use, VM use, and non-DS use by demographic, socioeconomic, and lifestyle characteristics. Prevalence of NVNM use in children was inclusive of those children taking only NVNM and children taking both VM and NVNM. Thus, VM only comprised those children who solely consumed VM. Descriptive statistics were generated as means or percentages, with a Taylor Series Linearization approach to approximate standard errors (SEs) for all estimates. These characteristics were compared by pairwise t-tests for categorical variables and orthogonal polynomial contrasts for ordinal variables as recommended by the NCHS (Table I).¹⁸ To evaluate trends in supplement use over time, P for trend across survey cycles was calculated using linear trend tests, taking into account the survey design (Table 2). Only those NVNM with a prevalence of ≥0.5% among all children at any time during the period (1999–2016) were analyzed. These trends were examined among all U.S. children and among all DS users. One-sided Rao-Scott Chi-square tests were used to test differences in NVNM use among DS users by age, sex, race/Hispanic origin, and family income over time (Figure 1), in motivations for use of NVNM and VM (Table 3), and in prevalence of each NVNM product type by age (Figure 2; available at www.jpeds.com). Motivations for use were also described for each NVNM product type (Table 5; available at www.jpeds.com). As recommended by the NCHS, estimates with a relative standard error ≥ 30% may be statistically unreliable, and these values were denoted in the tables;¹⁸ those with a relative standard error ≥40% were not reported.

Table 1.

Prevalence (% (SE))^a of use and non-use of DS in the past 30 days by product type and demographic, socioeconomic, and lifestyle characteristics among all U.S. children (aged ≤19y), NHANES 1999–2016

	n	Prevalence (SE) of NVNM useb	n	Prevalence (SE) of VM only use	n	Prevalence (SE) of no DS use
Percentage of all children (n=42510)	1032	3.6 (0.2)	9829	28.1 (0.5)	31649	68.3 (0.6)
Sex
Males	574	3.9 (0.2) 1	4822	27.3 (0.6) 1	16153	68.8 (0.7)
Females	458	3.3 (0.3) 2	5007	28.9 (0.7) 2	15496	68.8 (0.7)
Age, yearsb
<1	26	1.1 (0.2)	452	12.7 (0.7)	3667	86.2 (0.8)
1–5	212	2.5 (0.3)	3567	38.4 (0.9)	7439	59.1 (0.9)
6–11	282	3.7 (0.4)	2956	31.9 (0.8)	7835	64.4 (0.9)
12–19	512	4.5 (0.3)*	2854	20.8 (0.6)*	12708	74.7 (0.7)*
Race/ethnicityc
Non-Hispanic white	479	4.6 (0.3) 1	3738	33.2 (0.8) 1	7682	62.2 (0.8) 1
Non-Hispanic black	108	1.1 (0.1) 2	1946	17.7 (0.6) 2	9243	81.2 (0.7) 2
Hispanic	285	2.2 (0.2) 3	2995	20.4 (0.6) 3	12405	77.4 (0.7) 3
Family Incomed
Low	311	2.5 (0.3)	3664	20.0 (0.6)	17434	77.5 (0.7)
Middle	385	3.6 (0.3)	3455	29.7 (0.8)	9645	66.7 (0.8)
High	336	5.0 (0.4)*	2710	37.3 (0.9)*	4570	57.6 (0.9)*
Household’s Education Level
Less than high school	135	1.4 (0.2)	1748	15.9 (0.7)	11004	82.7 (0.8)
High school grad or equivalent	193	3.0 (0.4)	2119	24.7 (0.9)	7470	72.2 (0.9)
Some college/ associate degree	302	3.6 (0.3)	2948	30.4 (0.9)	7657	66.0 (0.9)
College degree or above	349	6.0 (0.5)*	2629	39.2 (1.1)*	4183	54.9 (1.1)*
Health Insurance
Private	461	5.2 (0.4) 1	3477	34.3 (0.8) 1	6882	60.5 (0.8) 1
Public	220	2.6 (0.3) 2	2285	20.3 (0.7) 2	10226	76.6 (1.6) 2
None	89	4.0 (0.8) 1,2	496	19.4 (1.3) 2	2366	74.7 (0.7) 2
Weight Status (≥2 years)
Underweight	33	3.3 (0.7)	351	35.9 (2.1)	698	60.7 (2.0)
Healthy	606	4.2 (0.3)	5657	31.4 (0.7)	14770	64.5 (0.7)
Overweight	141	3.6 (0.4)	1160	26.7 (1.0)	3935	69.7 (1.0)
Obese	131	3.1 (0.3)	1065	21.5 (0.8)*	5031	75.4 (0.9)*
Screen Time (≥2 years)
≤1 hour/day	109	3.8 (0.6)	1033	33.3 (1.6)	2722	62.9 (1.5)
>1–2 hours/day	187	4.8 (0.6)	1482	34.1 (1.1)	3338	61.2 (1.2)
>2–4 hours/day	277	3.9 (0.3)	2653	31.4 (0.8)	6768	64.7 (0.8)
>4 hours/day	228	3.2 (0.4)	2627	27.7 (0.8)*	8331	69.1 (0.8)*

DS, dietary supplement; NHANES, National Health and Nutrition Examination Survey; NVNM, non-vitamin, non-mineral DS; VM, vitamin or mineral containing DS

^aPercentage (SE, standard error) adjusted for survey weights of NHANES; not all categories have the same sample sizes either due to missing data or because data were not collected in those <2 y of age.

^bEstimates with different numbered subscripts (i.e., 1, 2, or 3) were significantly different across subgroups within each category at p <0.0001 except NVNM use by sex and VM use by sex, for which p <0.05;

asterisk (*) indicates significant linear trend at p<0.0001, except VM use by screen time, for which p<0.001. Children <1y of age were not included in the linear trend analyses for age. All χ² comparisons between user groups were statistically significant for all characteristics at p<0.0001, except sex for which p<0.01; none of the estimates had a relative SE ≥30%.

^cHispanic includes those who identified as Mexican American. The “other” racial category is not presented herein but is represented in the overall prevalence estimates.

^dPoverty–income ratio (PIR) was used as a proxy for family income; Low = PIR <130%, Middle = 130%≤ PIR ≤ 350%, and High = PIR >350%.

Table 2.

Trends in prevalence (% (SE))^a of all DS, NVNM, and VM use in the past 30 days among all U.S. children (≤19y of age) and of NVNM and VM use among children using any DS, NHANES 1999–2016^a

Among all children	Overall (n=42,517)	1999–2000 (n=5076)	2001–2002 (n=5621)	2003–2004 (n=5072)	2005–2006 (n=5367)	2007–2008 (n=4213)	2009–2010 (n=4316)	2011–2012 (n=4195)	2013–2014 (n=4405)	2015–2016 (n=4252)	Ptrendb
All DSc	31.7 (0.6)	30.0 (1.0)	33.0 (2.1)	30.8 (2.1)	31.3 (1.1)	29.3 (1.0)	32.0 (1.4)	31.5 (1.8)	33.1 (1.4)	34.1 (2.6)	0.20
VM supplements	28.1 (0.5)	27.1 (0.9)	30.6 (1.9)	28.3 (1.9)	28.6 (1.0)	26.9 (1.9)	28.4 (1.1)	27.3 (1.6)	27.9 (1.3)	27.7 (1.9)	0.54
NVNM supplements	3.6 (0.2)	2.9 (0.6)	2.4 (0.5)	2.4 (0.4)	2.8 (0.4)	2.4 (0.5)	3.6 (0.6)	4.2 (0.6)	5.1 (0.7)	6.4 (1.1)	<0.0001
Among all DS users	Overall (n=10,851)	1999–2000 (n=1181)	2001–2002 (n=1439)	2003–2004 (n=1162)	2005–2006 (n=1283)	2007–2008 (n=947)	2009–2010 (n=1170)	2011–2012 (n=1188)	2013–2014 (n=1265)	2015–2016 (n=1226)	Ptrendb
VM supplements	88.7 (0.6)	90.4 (2.0)	92.8 (1.3)	92.2 (1.3)	91.1 (1.0)	91.7 (1.6)	88.7 (1.6)	86.7 (1.5)	84.5 (2.0)	81.2 (2.2)	<0.0001
NVNM supplements	11.3 (0.6)	9.6 (2.0)	7.2 (1.3)	7.8 (1.3)	8.9 (1.0)	8.3 (1.6)	11.3 (1.6)	13.3 (1.5)	15.5 (2.0)	18.8 (2.2)	<0.0001
Omega-3d	1.2 (0.1)	-	-	-	2.1 (0.6)	3.2 (1.2)†	5.6 (1.4)	5.5 (1.1)	6.0 (1.1)	6.7 (1.5)	<0.0001
Probiotics	0.4 (0.1)	-	-	-	-	-	-	-	1.5 (0.4)	3.3 (0.8)	<0.001
Fiber	0.3 (0.0)	-	-	-	-	2.2 (0.6)†	-	1.0 (0.4)†	2.1 (0.6)	3.4 (0.5)	<0.0001
Melatonin	0.3 (0.0)	-	-	-	-	-	-	2.2 (0.7)†	2.7 (0.9)†	3.3 (0.6)	<0.0001
Botanicalse	0.6 (0.1)	2.9 (0.7)	2.6 (0.5)	2.7 (0.9)†	1.8 (0.7)†	1.3 (0.5)†	-	-	2.7 (0.6)	1.8 (0.6)†	0.17
Echinaceaf	1.3 (0.2)	-	1.6 (0.4)	1.8 (0.6)†	-	-	0.7 (0.2)	-	1.3 (0.3)	-	0.19

DS, dietary supplement; NVNM, non-vitamin, non-mineral DS; VM, vitamin or mineral containing DS; NHANES, National Health and Nutrition Examination Survey

^aPercentage (SE, standard error) adjusted for survey weights of NHANES

^bA Bonferroni-corrected p-value of 0.0056 was considered to be statistically significant

^cIncluding VM and NVNM

^dOmega-3 PUFAs (polyunsaturated fatty acids)

^eMain/only ingredient is a single plant-based supplement

^fNot mutually exclusive of VM or other NVNM

^†Relative SE ≥30%; estimates with relative SE ≥40% not reported

Figure 1.

Trends in prevalence (% (SE)) of NVNM supplement use among U.S. children (≤19y of age) using any DS by sex, age, race/Hispanic origin, and family income, NHANES 1999–2016

NVNM, non-vitamin, non-mineral DS; DS, dietary supplement; NHANES, National Health and Nutrition Examination Survey; prevalence percentage (SE, standard error) adjusted for survey weights of NHANES

a. Prevalence of any NVNM use among DS users by sex; b. Prevalence of any NVNM use among DS users by age; prevalence among <1y not presented due to unreliable estimates; c. Prevalence of any NVNM use among DS users by race/Hispanic origin; d. Prevalence of any NVNM use among DS users by family income (poverty–income ratio (PIR) was used as a proxy for family income; Low = PIR<130%, Middle = 130%≤PIR≤350%, and High = PIR>350%); *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001

Table 3.

Prevalence (% (SE))^a of reported motivations for use of NVNM and VM in the past 30 days among U.S. children (≤19 y of age) using any DS, NHANES 2007–2016

Motivation	NVNM users (n=785)	VM users (n=6,454)	p-valueb
To maintain health (stay healthy)	23.1 (2.7)	38.7 (1.3)	<0.0001
To improve overall health	22.6 (2.3)	33.1 (1.4)	<0.0001
To prevent health problems	13.1 (1.9)	14.4 (1.0)	0.52
For relaxation, stress, sleep	12.7 (1.9)	0.8 (0.3)†	<0.0001
To boost immunity, prevent colds	10.9 (1.6)	16.5 (1.1)	0.01
To supplement the diet	10.5 (2.1)	20.6 (1.4)	<0.0001
For mental health	10.4 (1.8)	1.3 (0.3)	<0.0001
For bowel/colon health	10.1 (1.7)	0.7 (0.2)	<0.0001
To improve digestion	7.3 (1.3)	1.2 (0.2)	<0.001
For heart health, cholesterol	5.4 (1.2)	0.8 (0.2)	<0.0001
For healthy skin, hair, and nailsc	4.8 (0.8)	3.2 (0.4)	0.046
To get more energy	4.5 (0.9)	5.3 (0.5)	0.39
For bone health	2.7 (1.1)	5.6 (0.8)	0.01
For weight loss	2.6 (0.8)†	-	<0.0001
Other reasons	2.6 (0.6)	1.5 (0.3)	0.09
For healthy joints, arthritis	2.5 (0.7)	1.0 (0.2)	<0.01
To build muscle/weight gaind (n=700)	2.3 (0.7)†	1.4 (0.2)	0.05
For eye health	1.0 (0.3)†	1.0 (0.2)	0.91
For teeth, prevent cavities	-	3.2 (0.5)	<0.0001
For anemia, such as low iron	-	2.4 (0.3)	<0.0001
To maintain blood sugar, diabetes	-	1.3 (0.2)	0.35
Kidney and bladder health	-	0.4 (0.1)	0.79

NVNM, non-vitamin, non-mineral DS; VM, vitamin or mineral containing DS; DS, dietary supplement; NHANES, National Health and Nutrition Examination Survey

^aPercentage (SE, standard error) adjusted for survey weights of NHANES; participants were able to select more than one motivation for each product

^bp-value based on Rao-Scott chi-square test

^cHair and nails were in a separate category in 2007–2010 and combined in 2011–2016

^dCategory not included in 2007–2008; combined in 2009–2010; included only weight gain in 2011–2012; and included as separate questions in 2013–2016 (but combined for this analysis)

^†Relative SE≥30%; estimates with relative SE≥40% not reported

Figure 2 online.

Prevalence (% (SE))^a of use for each NVNM product type in the past 30 days among U.S. children (≤19y of age) using any NVNM DS by age, NHANES 1999–2016

NVNM, non-vitamin, non-mineral DS; DS, dietary supplement; NHANES, National Health and Nutrition Examination Survey

^aPercentage (SE, standard error) adjusted for survey weights of NHANES; prevalence of echinacea too small to stratify

*Significant difference by age group at p<0.0001 using Rao-Scott chi-square test; the chi^-square test for melatonin only included the last three age groups because the prevalence of melatonin among <1y was 0%

Table 5 online.

Prevalence (% (SE))^a of specific NVNM products and their most frequently reported motivation for use in the past 30 days among U.S. children (≤19 y of age) using any NVNM DS, NHANES 2007–2016

	Prevalence (SE) among NVNM users (n=785)	Most commonly reported motivation	Prevalence (SE) of motivation
Omega-3 PUFAs	34.8 (2.5)	To maintain health (stay healthy)	35.5 (3.8)
Probiotics	10.9 (1.6)	For bowel/colon health	41.8 (4.0)
Melatonin	10.9 (1.6)	For relaxation, stress, sleep	90.3 (2.0)
Botanicalsb	10.8 (1.5)	To improve overall health	38.5 (5.4)
Fiber	9.7 (1.1)	For bowel/colon health	49.4 (5.5)
Echinaceac	2.6 (0.9)†	To boost immunity, prevent colds	55.8 (14.9)

NVNM, non-vitamin, non-mineral DS; DS, dietary supplement; NHANES, National Health and Nutrition Examination Survey; PUFA, polyunsaturated fatty acid

^aPercentage (SE, standard error) adjusted for survey weights of NHANES; participants were able to select more than one motivation for each product

^bMain/only ingredient is a single plant-based supplement

^cNot mutually exclusive of VM or other NVNM

^†Relative SE≥30%

Results

Descriptive Characteristics

As nationally representative of U.S. children, most (~68%) do not use any DS; about 4% of all children reported NVNM use whereas 28% reported VM use from 1999–2016 (Table 1). Prevalence of NVNM use was significantly higher in boys when compared with girls (3.9 vs 3.3%, p<0.05), and increased with age, regardless of sex; approximately 1% of infants reported DS use, and 4.5% of adolescents 12–19y reported doing so (p_trend<0.0001). NVNM use significantly differed by race/Hispanic origin groups; although only ~1% of non-Hispanic Black children reported taking an NVNM, use of NVNM products among Hispanic children was 2.2% and 4.6% among non-Hispanic White children (p<0.05). NVNM use was also linearly associated with family income (p_trend<0.0001) and household education (p_trend<0.0001). Children living in households with an education level of a college degree or higher had the highest use of NVNM (6.0%) whereas children living in households with an education level of less than high school had the lowest use (1.4%). NVNM use also tended to differ by type of health insurance; children with private health insurance (5.2%) were significantly more likely to use NVNM when compared with children who were insured under public health insurance (2.6%, P < .05). However, neither of these groups of children differed in prevalence of NVNM use when compared with those with no health insurance (4.0%). NVNM use did not follow a trend when examined by children’s weight status or amount of time spent in front of a screen. However, using pairwise comparisons, NVNM use was significantly lower among children with obese weight status compared with those with normal weight status (p<0.05, data not shown). Furthermore, NVNM use was significantly lower among children who spent >4 hours/day in front of a screen when compared with those who had >1–2 hours/day of screen time (p<0.05, data not shown).

Between 1999 and 2016, VM use was reported by 28% of all U.S. children and significantly differed by all descriptive characteristics (Table 1). Opposite the patterns observed with NVNM use, VM use was significantly higher among girls when compared with boys (28.9 vs 27.3%, p<0.05) and decreased by age (when infants <1y were excluded from analysis); the highest prevalence of reported VM use was among the youngest children (1–5y; 38.4%) (p_trend<0.0001). The pattern of VM use by race/Hispanic origin was similar to that for NVNM use (p<0.05). VM use also followed similar trends by family income and household education level when compared with NVNM use (p_trend<0.0001 for each). VM use was higher among children with private insurance when compared with children with public or no insurance, who did not differ. Prevalence of VM use decreased by weight status among children; nearly 36% of underweight children consumed VM, and only 21.5% of children with obesity did so (p_trend<0.0001). The prevalence of VM use also decreased by hours of screen time (p_trend<0.0001). Patterns of prevalence for no DS use for each of these characteristics were in the opposite direction as those for prevalence of VM use. However, there was no difference by sex.

Trends in Prevalence

Use of any DS and VM among all children remained relatively stable over time, with around 30% reporting use between 1999–2000 and 2015–2016 (Table 2). On the contrary, use of NVNM among all children increased by 120% during this period from 2.9% in 1999–2000 to 6.4% in 2015–2016 (p_trend<0.0001). NVNM use among children using any DS also increased by 96%, from 9.6% in 1999–2000 to 18.8% in 2015–2016 (p_trend<0.0001). Children using only VM made up the remaining proportion of DS users, and thus VM had a significant decreasing trend (p_trend<0.0001).

In the earlier survey years, with the exception of single botanicals and echinacea, NVNM use was low regardless of product type (Table 2). Omega-3 PUFA products increased by over 200%, from 2.2% in 2005–2006 to 6.8% in 2015–2016 (p_trend<0.0001). Prevalence of probiotic, fiber, and melatonin DS use was negligible in the previous decade, yet increasing trends in use were observed beginning in 2011 until 2016 (p_trend<0.0001 for all). From 2013–2016, use of probiotics increased by 120%, and use of fiber increased by 240%. Melatonin also increased, though not as rapidly as other product categories (22% increase). Use of single botanicals and echinacea did not show any discernable trend over this period. Other common NVNM that were used (but <0.5% of all children), included garlic, elderberry, and amino acids DS (data not shown).

Across the study period (1999–2016), there was no significant difference in NVNM among all DS users by sex until 2015–2016, when boys had higher prevalence of NVNM use compared with girls (p<0.001, Figure 1, A). Older children (12–19y) consistently had higher prevalence of NVNM use than younger children (Figure 1, B). However, the younger age groups (1–5y and 6–11y) exhibited a steeper increase about halfway through the study period, narrowing the gap by 2015–16. Among specific NVNM products, the prevalence of probiotic and botanical use differed by age of the child (Figure 2). No differences were found for specific NVNM product types by sex of the child (data not shown).

Non-Hispanic White children had the highest prevalence of NVNM use compared with Hispanic and non-Hispanic Black children (Figure 1, C). In general, marginal differences in the prevalence of NVNM use were observed between Hispanic and non-Hispanic Black children over time, with a notable divergence in 2007–2008. After a sharp decrease among non-Hispanic Black children during this time (p<0.01), their NVNM use slowly recovered after 2011–2012. Children living in high-income households tended to have higher NVNM use than those in low- and middle-income households, and this difference became significant in 2015–2016 (Figure 1, D).

Motivations

The two most common motivations for DS use were “to maintain health” and “to improve overall health”, though the percentages among VM were higher than those for NVNM (Table 3). For children under 2y of age, the top two motivations were “to improve digestion” and “to improve overall health” ( data not shown). More children living in low-income households reported “to improve overall health” compared with those in middle-income households (p<0.001) and in high-income households (p<0.05) (data not shown). Motivations for each type of DS product are reported in Table 5.

Less than 20% of participants reported taking NVNM at the recommendation of a physician or healthcare provider; however, this percentage did not differ for VM (data not shown). Notably, there was usage, albeit sparse, of some botanicals with safety concerns, including St. John’s wort, kava kava, and dong quai (data not shown).

Discussion

In this nationally representative analysis of U.S. children over the past two decades, there was an increase in the reported use of NVNM and overall DS use and VM use remained stable. Although DS use among U.S. adults trended upwards from 1999–2006,⁴ a more recent study showed overall DS and multivitamin multi-mineral use to be stable from 1999–2012, consistent with our findings in children; however, trends in NVNM use varied by product type.⁵ Our data also confirm previous papers with regard to patterns of VM use by demographic characteristics of DS users, including race/Hispanic origin, family income, and household educational attainment, as well as motivations for their use.^{3, 6, 11} However, our findings of higher prevalence of NVNM use among boys and of the opposite pattern for VM use is novel. No previous research has identified differential patterns of DS use by sex among children for VM,^{3, 6} or for NVNM intake.^{20, 21} We also found that NVNM use appears to be higher in adolescents than in younger children, which is contrary to the patterns observed for VM products. Our findings that NVNM use is highest among older children, while VM use is highest among younger children, is consistent with previous studies.^{3, 6, 11, 20}. However, patterns of NVNM and VM use were more consistent by race/Hispanic origin; NVNM use was highest among non-Hispanic White children when compared with non-Hispanic Black and Hispanic children, a pattern that has been previously observed with VM supplements.^{22, 23} Similar to race/Hispanic origin, trends in VM use observed by weight status,^{3, 6, 24, 25} screen time,^{3, 6} and health insurance^{3, 6, 11} were also supported by previous literature evaluating NHANES data in children. NVNM use by type of health insurance also followed a previously reported pattern.^{11, 20} No studies were found reporting NVNM use by weight status or screen time.

The increase in use of NVNM across time appears to have been largely driven by the use of omega-3 PUFAs, probiotics, fiber, and melatonin. Our findings closely match that of a recent study that examined DS use and trends in children, including prevalence of omega-3 PUFAs, melatonin, and overall NVNM supplement use.²⁶ However, where they used data from 2013–2014 for their main analyses, our study spans almost two decades, providing much more information and context.

In our study, omega-3 PUFAs use did not differ across age groups and were primarily reported to be used to maintain health. This is in line with previous reports of increasing omega-3 PUFA consumption among U.S. adults.^{2, 5} Though little is known about omega-3 supplementation patterns in children, some data suggest children (1–19y) have significantly lower energy-adjusted intakes of omega-3 PUFAs from the diet when compared with adults.²⁷ Clinical trials with omega-3 supplementation in children generally focus on mental health and behavioral disorders; although the evidence is inconclusive, several such trials have reported therapeutic benefits.^28–34 Omega-3 supplementation in early childhood has also been suggested to improve psychomotor and visual development,³⁴ modify effects of indoor air pollution in children with asthma,³⁵ and be effective in improving cardiovascular conditions in some children.^{36, 37} In adults, however, associations of omega-3 PUFA supplementation with most health outcomes have largely been null.^{38, 39}

Our results also show a trend of increased use of probiotics and fiber, chiefly for bowel and colon health. Both product types were used more frequently in younger children than in adolescents. The prevalence of these products is still relatively small and could only be reliably estimated for the most recent survey cycles. This is in line with the trajectory of research on probiotics and prebiotic fibers, which has expanded in the previous decade.⁴⁰ The gut microbiome has been shown to modulate the immune system as well as neurodevelopment via biochemical signaling.⁴¹ Studies have found beneficial effects of specific probiotic strains for gastrointestinal conditions, allergies, major depressive disorders, and autism spectrum disorders, among others.^40–44 However, the evidence is not conclusive, and larger well-designed randomized control clinical trials are needed to validate these potential benefits.

Our findings also demonstrate an increase in the use of melatonin mainly for relaxation, stress, or as a sleep aid. Previous studies suggest melatonin is effective in treating children with insomnia.^45–47 A meta-analysis found a significant increase in sleep time among children with neurocognitive disorders when given melatonin.⁴⁸ Another meta-analysis reported that melatonin is overall well tolerated; however, long-term safety of the supplement has yet to be examined.⁴⁹ U.S. sales of sleep supplements have been increasing steadily over the last three years as a whole; sales increased by 17.4% in 2019, and are projected to increase by over 30% in 2020, with a 17.1% increase in melatonin alone.⁵⁰

Though there is still conflicting evidence of its benefits, echinacea has long been purported for enhancing immunomodulatory properties.^{20, 51} Recent meta-analyses of clinical trials have shown effectiveness of echinacea for treating upper respiratory tract infections in young children by itself⁵² and by enhancing the efficacy of vitamin C.⁵³ However, not all of the trials in these meta-analyses showed a significant improvement, and some showed either no effect or adverse effects. Additional studies are warranted for examining the effects of echinacea alone and in conjunction with vitamin C.

The use of some DS among children and adolescents has been associated with adverse events,^{54, 55} and may have the potential to interact with prescription medications.⁵⁶ At the same time, there have been observed benefits of NVNM in conjunction with medications, such as melatonin for insomnia treatment among children with ADHD using stimulant types of prescription medications.⁵⁷

Strengths and Limitations

Our study leveraged almost two decades of data from NHANES, providing a large nationally representative sample to describe usage trends of DS among U.S. children. Trained researchers carefully classified all DS products into mutually exclusive categories, which have been consistently reported previously in the literature.^{2, 3, 10, 19} However, this was a manual process, and subjective decisions were made. As differential classifications may lead to significant variability in prevalence estimates of specific products, there is need for a standardized classification system for DS, especially NVNM.⁸ Another potential limitation was the use of proxies for children under 16 years of age as the reporter for DS used, as information from caregivers may be subject to bias, such as recall or social desirability bias. However, very little is known about measurement error in DS reporting.⁵⁸ Furthermore, NHANES only began oversampling non-Hispanic Asians starting in 2011–2012; thus, we were not able to include this race/ Hispanic origin category in our analysis. Lastly, several NVNM prevalence estimates were too low to report, especially for earlier survey years, and many had unreliable standard errors. However, by 2015–16 most estimates were considered reliable.

Despite increases in the prevalence of use of NVNM products, high quality evidence supporting their use is lacking, especially in children. Our work presented here serves to provide nationally representative trend data on NVNM (and VM) and motivations for their use. More rigorous studies, such as randomized controlled trials, are needed to inform healthcare practitioners, children, and caregivers on the safety and effectiveness of NVNM.

Abbreviations:

BMI

Body mass index

CDC

Centers for Disease Control and Prevention

NCHS

National Center for Health Statistics

NHANES

National Health and Nutrition Examination Survey

NVNM

Non-vitamin/non-mineral containing dietary supplements

PIR

Income-to-poverty ratio

PUFA

Polyunsaturated fatty acid