Follow us on :



Take a look at the Recent articles

Blood inorganic mercury is directly associated with glucose levels in the human population and may be linked to processed food intake

Renee Dufault

Food Ingredient and Health Research Institute, Naalehu, Hawaii, USA

Fort Peck Community College, Poplar, Montana, USA

E-mail : rdufault@foodingredient.info

Zara Berg

Fort Peck Community College, Poplar, Montana, USA

Raquel Crider

Food Ingredient and Health Research Institute, Naalehu, Hawaii, USA

Shepherd University, Shepherdstown, West Virginia, USA

Roseanne Schnoll

Food Ingredient and Health Research Institute, Naalehu, Hawaii, USA

Department of Health and Nutrition Sciences, Brooklyn College of City University of New York, Brooklyn, New York, USA

Larry Wetsit

Fort Peck Community College, Poplar, Montana, USA

Wayne Two Bulls

Fort Peck Community College, Poplar, Montana, USA

Steven G. Gilbert

Food Ingredient and Health Research Institute, Naalehu, Hawaii, USA

Institute of Neurotoxicology and Neurological Disorders, Seattle, Washington, USA

H.M. “Skip” Kingston

Department of Chemistry and Biochemistry, Duquesne University, Pittsburgh, Pennsylvania, USA

Mesay Mulugeta Wolle

Department of Chemistry and Biochemistry, Duquesne University, Pittsburgh, Pennsylvania, USA

G.M. Mizanur Rahman

Department of Chemistry and Biochemistry, Duquesne University, Pittsburgh, Pennsylvania, USA

Dan R. Laks

Department of Biological Chemistry, University of California Los Angeles (UCLA), Los Angeles, California

DOI: 10.15761/IMM.1000134

Article
Article Info
Author Info
Figures & Data

Abstract

Background

The goals of the study were (1) to determine the impact of inorganic mercury exposure on glucose homeostasis; and (2) to evaluate the effectiveness of two community-based interventions in promoting dietary changes among American Indian college students to reduce risk factors for Type-2 Diabetes including fasting glucose, insulin, and mercury levels, weight, and body mass index.

Methods

To accomplish goal one, the National Health and Nutrition Examination Survey (NHANES) dataset was analyzed using a previously published method to determine if there is a relationship between inorganic blood mercury and fasting glucose. To accomplish goal two, ten college students were recruited and randomly assigned to a group receiving the online macroepigenetics nutrition course and the support group for eliminating corn sweeteners. Participants in both groups were assessed for diet patterns, weight, body mass index (BMI), fasting glucose, insulin, and mercury levels. The interventions were implemented over a 10-week period.

Results

Analysis of the NHANES data (n=16,232) determined a direct relationship between inorganic mercury in blood and fasting glucose levels (p<0.001). The participants who took the online macroepigenetics nutrition intervention course significantly improved their diets (p<0 .01), and fasting blood glucose levels (p<0.01) while having lower levels of inorganic mercury in their blood compared to the subjects in the group who eliminated corn sweeteners from their diet and participated in the support group. The trend in lower blood inorganic mercury was strong with p=0.052. The participants in the support group who eliminated corn sweeteners from their diet achieved significant weight loss (p<0.01) and reduced their body mass index (p<0.01).

Conclusion

Total blood mercury levels may be influenced by dietary intake of highly processed foods and lower inorganic mercury levels are associated with lower fasting glucose levels. Alternative community-based interventions emphasizing the role food ingredients and toxic substances play in gene modulation and the development of diseases can result in significant dietary improvements and reductions in risk factors associated with type-2 diabetes. A healthier diet can be promoted among community members using a novel online nutrition course. Consumption of corn sweeteners may be a risk factor in the development of obesity.

Key words

macroepigenetics, fructose, corn syrup, diet, diabetes, online, glucose, NHANES, mercury

Background

American Indians and Alaska Natives are persons having origins in any of the indigenous populations of North America who maintain tribal affiliation or membership [1]. Indian Country is comprised of American Indian (AI) and Alaska Native (AN) communities across the United States of America (USA).  Compared with other Americans, AI and AN populations experience a disproportionate burden of chronic disease including liver disease and cirrhosis, diabetes, and heart disease [2,3]. Heart disease and type-2 diabetes (T2D) are the top two leading causes of death in AI and AN communities [2]. Diabetes prevalence rates are increasing among the AI and AN population aged 20 years or older and vary by region from 6.0% among AN adults to 24.1% among AI adults in southern Arizona [4,5]. Obesity is a significant risk factor in the development of both heart disease and T2D and contributes to the high prevalence rates of these diseases in Indian Country [6]. Although excess caloric consumption and a sedentary lifestyle are well known risk factors for obesity, T2D and heart disease, there is increasing evidence to suggest that exposure to toxic environmental substances and depletion of dietary micronutrients may play a significant role in the etiology of these diseases [7-10]. Toxic metal and pesticide exposures and excess consumption of food ingredients known to affect micronutrient status have been linked to the development of diabetes [8-11] through gene-environment interactions [8,10,11].

Adequate dietary intake of micronutrients is necessary to sustain metabolism and tissue function [12]. A recent study reported that concentrations of chromium (Cr), copper (Cu), manganese (Mn), nickel (Ni), lead (Pb) and zinc (Zn) in foodstuffs significantly correlated with that in human blood following intake [13].  In the only published report thus far about the dietary intake of a representative sample of the AI population, many of the AI men and women were not meeting the dietary recommendations for the key micronutrients magnesium (Mg) and Zn [14]. This finding supports the conclusion made a few years earlier by USA Department of Agriculture scientists who reported nearly one half of all Americans one year old and over had inadequate intakes of dietary Mg [15]. Such micronutrient insufficiencies over time may lead to increased risk for heart disease, diabetes, and neurodevelopmental disorders [16-19].  Zn, Mg, and phosphorus (P) losses can occur as a result of excess consumption of high fructose corn syrup (HFCS) and create imbalances in dietary micronutrient status [20,21]. Such imbalances are problematic from a macroscopic gene-environment point of view since dietary micronutrient status can either exacerbate or mitigate the effects of exposure to toxic environmental substances by altering gene function especially in the case of child neurodevelopment and diabetes [17,22,23].  Fructose consumption in dietary conditions of magnesium deficiency induces insulin resistance while lowering PON1 gene activity [22,23].  PON1 gene expression is needed by the body to metabolize organophosphate pesticides known to adversely impact child neurodevelopment [17]. Child and reproductive health and diabetes are all health issues of grave concern to the indigenous populations of North America [24].

Health education can play a role in addressing these health issues if it is delivered in a culturally competent manner. Community based and culturally competent education efforts have made a positive difference in diabetes prevention efforts in minority communities by empowering community members to make lifestyle changes through the acquisition and dissemination of diabetes knowledge [25]. During the previous year, Fort Peck Community College (FPCC) collaborators developed and evaluated the efficacy of a culturally competent online nutrition intervention course [26]. The course was found to be effective in producing healthful dietary changes among community members who completed it [26].

  The curriculum of the course is based on the underlying assumption that the prevalence of T2D among the AI population is due to the industrialization of the food supply, dietary exposure to toxic substances, and a general lack of knowledge among community members as to the epigenetic role food ingredients, nutrition, and invasive toxic substances play in the development of diseases. Epigenetic changes impacting human metabolism and health can occur through nutrition via dietary intake of methyl donating nutrients such as choline and betaine [27-29].  Macroepigenetics is a theoretical, consumer friendly approach that allows laypeople to consider how factors of nutrition, environment and gene expression interact to contribute to the development or prevention and inheritance of disease [17].  In accessing the macroepigenetics nutrition intervention course, participants are provided numerous opportunities to review research on the role nutritional factors and invasive toxic substances in the food supply play in gene modulation making them more susceptible to diabetes and other disease conditions. Participants learn that genes turn on and off in response to diet to produce the hormones and proteins needed by the body to regulate metabolism. They also learn that exposure to toxic substances such as mercury and fructose interfere with body metabolism. Participants finishing the online nutrition intervention course earn 3 units of science credit at FPCC. In this study, we wanted to determine if dietary changes resulting from participating in the online nutrition intervention course could result in reductions in risk factors associated with T2D including fasting blood glucose, insulin and mercury levels, weight, and body-mass-index (BMI).

Recognizing a one size fits all approach rarely works to resolve any health issue; we also wanted to offer an alternative intervention for individuals wishing to improve their health status by eliminating their consumption of corn sweeteners. Synthesized from corn starch [30] the targeted corn sweeteners included high fructose corn syrup (HFCS), corn syrup, modified corn starch, dextrose, maltodextrose, maltodextrin, and fructose. From a macroepigenetic perspective, HFCS is an invasive toxic substance for at least three different reasons. Its consumption by humans can result in dietary mercury exposure [31, 32], and insulin resistance [33-35] and reduced PON1 gene activity in rats [22]. PON1 is the gene responsible for producing the paraoxonase enzyme required for breaking down the toxic organophosphate pesticide residues [22] found in wheat, corn, and wheat products [36].  Patients with T2D who have insulin resistance also have lower paraoxonase activity [37]. While it is unclear how decreased PON1 activity contributes to the development of T2D, there is evidence to suggest that inorganic mercury exposure plays a role in insulin resistance.  Inorganic mercury exposure in the mercuric chloride (HgCl2) form is suspected to be a contributing factor to the onset of insulin resistance by interfering with genes that regulate glucose homeostasis [38,39]. Exposure to inorganic mercury in the environment has also been found to contribute to the development of insulin resistance in non-diabetic humans by interacting with other toxins, such as dioxin [28,40].

  Mercury exposure can be determined through the analysis of a variety of tissues to include blood, urine, finger or toe nails, breast milk and hair [41]. Many studies have measured total Hg in blood without distinguishing the forms of mercury found in the blood [41]. This study was focused on determining the form of mercury exposure from consumption of processed foods.  Inorganic mercury may enter food products during the various manufacturing processes.  For example, mercury cell chlor-alkali chemical products are used extensively in food processing and always contain inorganic mercury residues. Vegetable oil products manufactured using the common alkali refining process may present a moderate risk of mercury contamination [42]. The mercury cell chlorine used to bleach flour is expected to contain a small amount of mercury residue [43]. The corn starch used to manufacture the corn sweeteners in the HFCS product line is treated purposely with inorganic HgCl2 as part of the manufacturing process to inhibit endogenous starch-degrading enzymes [30]. It is thus reasonable to suggest that consumers are routinely exposed to non-elemental inorganic mercury (I-Hg) when they consume heavily processed foods, including corn sweeteners.  Our justification for adding the support group intervention to help students and community members eliminate corn sweeteners from their diet is based on the concept that consumption of corn sweeteners is both a known and potential source of inorganic mercury exposure [31,32] and a potential factor in the development of insulin resistance [44].

Methods

The first goal of this study was to determine the impact of inorganic mercury exposure on glucose homeostasis through an analysis of the NHANES dataset.

NHANES dataset analysis

The NHANES is a dataset of measured outcome variables such as biomarkers for the target population of the non-institutionalized, civilian United States population. The NHANES survey is generated in data groups of separate 2-year clusters. Our study analyzed the NHANES datasets combining survey clusters from 1999-2012 and was focused on the following biomarkers: fasting glucose, blood inorganic mercury (I-Hg), organic mercury, and insulin. In addition to analyzing blood I-Hg as a continuous variable, we also assessed its associations as a binary variable.

The rationale and methods to generate the binary variable blood I-Hg detection (I-Hg detect) are detailed in Laks [45]. We set a constant value across survey years for the limit of detection (LOD) based on the first survey years limit of detection (0.4 ug/L) and used that as a standardized baseline.  Thereby, we generated a binary variable for I-Hg detection, where 0=no detection (below 0.4 ug/L) and 1=positive detection (above or equal to 0.4 ug/L). This maintained a standardized limit of detection for all survey groups. This coding procedure was followed because most values for blood I-Hg fell below the LOD and were assigned estimate values that changed across survey years. Thus, LBXIHG, the NHANES variable for I-Hg is not an optimal continuous variable but may be better assessed as a binary variable of detection/no detection. To compensate for the changing estimate values below the LOD, values below our assigned and uniform LOD of 0.4 ug/L were assigned a constant estimate value of 0.3 ug/L which was the estimate value used for the initial 1999-2000 survey set. We stress that this was done only to estimate the rate of I-Hg detection in blood. In order to determine associations with blood I-Hg concentration we also used the primary NHANES continuous variable, LBXIHG (blood I-Hg). We found comparable results for associations with both I-Hg detection and I-Hg mean concentration. In addition, we generated a value for blood organic mercury, CH3Hg by subtracting blood I-Hg (LBXIHG) from total blood mercury (LBXTHG).

This study analyzed both the raw population and the survey weighted population in order to detect robust associations. The survey-weighted population extends the inferences based on our findings to the U.S. population and utilizes a weighting adjustment as outlined in the NHANES website. The stratum and PSU variables assist in estimating variances in order to reflect the design structure of the NHANES survey. In STATA the 1999-2012 NHANES combined data set is weighted for survey analysis by: svyset(pw=WT99-2012), psu(sdmvpsu), strata(sdmvstra) where WT99-2012 is the combined weight calculated above. We chose to report as significant only those associations with p-values less than 0.01 in the survey-weighted population.

The NHANES population analyzed was the full NHANES population available with measured blood I-Hg levels.  However, we also adjusted for potential confounders and used multivariate analysis to adjust for age, race and gender.  We used linear regression and logistic regression models when appropriate.  Statistical analysis of the NHANES dataset was performed using STATA 8.0 (StatCorp).

Health education interventions

The second goal of this study was to evaluate the effectiveness of two health education interventions in reducing risk factors for T2D: the online macroepigenetics nutrition course (MAC) and the support group for corn sweetener elimination (CSE). Participants in both groups were assessed for weight, BMI, fasting glucose, insulin, and mercury levels.  The interventions were implemented over a 10-week period.

This study was conducted at the Fort Peck Community College (FPCC) in compliance with a protocol approved by a third party non-profit Institutional Review Board (IRB), the BioMed IRB [46].  The tribe did not have their own IRB at the time but the Fort Peck Tribal Executive Board passed a resolution in support of the members participating in a macroepigenetic study of changes in health status following the nutrition education intervention program at FPCC. Located in Popular, Montana on the Fort Peck Indian Reservation, FPCC is a tribally controlled community college chartered by the government of the Fort Peck Assiniboine and Sioux Tribes. The macroepigenetics nutrition intervention course (MAC) and pretest and posttest food frequency survey questionnaire were delivered on line via the non-profit Food Ingredient and Health Research Institute (FIHRI) website [47] to the FPCC students.  The corn sweetener elimination (CSE) support group intervention was delivered by FPCC teaching staff. While ten community members initially enrolled in the study only nine successfully participated in one of the alternative interventions offered during the study.

The nursing staff at the Northeast Montana Health Services (NEMHS) Riverside Clinic screened ten participants and randomly assigned half of them to the CSE intervention group.  The five participants in the CSE group were asked to eliminate corn sweeteners in the HFCS product line from their diet and participate in a support group. Screening was conducted to ensure that participants, regardless of the intervention assigned, were at least 20 years of age and not on any medication except birth control. The nursing staff was also responsible for collecting weight and height measurements, and blood samples for biomarker analysis. 

 Community members participating in either the online intervention course (MAC) or the CSE group received a monthly stipend of $200 for each month of successful participation. Success in the nutrition intervention course (MAC) was determined by completed homework submissions, participation in the online peer group discussion forum, and completion of a culminating final project. Success in the support group (CSE) was determined by meeting weekly with the support group coordinator.  The stipend was provided to compensate the community members for their time in completing the online course work with the embedded pre and post food frequency survey or meeting weekly with the support group coordinator, and providing blood samples for analysis and height and weight measurements for determining BMI.

MAC intervention course

The online MAC course consisted of ten modules of culturally competent instruction delivered over ten weeks at the FIHRI website which also provides an outline of the course content [47]. Each of the five participants in the course was provided with their own unique password to access the modules of instruction. The curriculum provided opportunities for independent research, collaboration, and peer group interactions to discuss lessons learned.  For example, during the third module of instruction, participants were required to access the United States Department of Agriculture (USDA) Food Availability Data System and determine the average per capita availability and consumption for various foods consumed in the US [48]. After compiling their data, they were then asked to interpret the changes in commodity consumption over time and how these changes might contribute to the development of modern diseases. Table 1 provides an example of the data the participants gathered for sugars and vegetable oils from 1970 to 2010. The participants were able to compare the per capita American consumption rates of high fructose corn syrup, cane and beet sugar, total corn sweeteners, and vegetable oils over time. Using the USDA data, the participants determined that from 1970 to 2010, cane and beet sugar consumption decreased 35% while high fructose corn syrup consumption increased 9,467%.  During the same period of time, per capita vegetable oil consumption increased 248%. In their online forum, they discussed the potential impact of this increased high fructose corn syrup and vegetable oil consumption on Indian health. In a later module, the participants reviewed peer reviewed journal articles and learned that fructose consumption impacts PON1 gene activity which may create conditions of oxidative stress leading to the development of insulin resistance and T2D [22].

Table 1. Change in American sugar and vegetable oil consumption 1970-2010

Commodity

1970 per capita consumption (Lbs/year)

2010 per capita consumption (Lbs/year)

Percent increase or decrease

Cane and beet sugar

59.8

38.7

- 35%

High fructose corn syrup

0.3

28.7

+9,467%

Total corn sweeteners, including high fructose corn syrup

9.3

37.8

+306%

Vegetable oils (salad and cooking)

15.4

53.6

+ 248%

To evaluate whether or not the curriculum was effective in reducing participant consumption of food commodities and ingredients that lead to the development of chronic diseases, a one-group pretest-posttest survey was designed and administered online prior to and after receiving the instruction over the 10-week long intervention period.  The IRB approved survey was constructed and delivered using the online Survey Monkey tool [49]. Food frequency questions were modeled after those used by the National Cancer Institute [50] to query dietary intake during the past month. Using the same format we developed three additional questions to determine the intake of organic flour, organic vegetables and fruit, and organic processed foods (crackers, bread, and cereal).

CSE support group intervention

Participants in the CSE support group were provided the following:

  • A shopping guide with food ingredients to avoid including HFCS, corn syrup, modified corn starch, dextrose, maltodextrose, maltodextrin, and fructose.
  • Instructions on the importance of reading food ingredient labels as opposed to “nutrition facts.”
  • Field trip to grocery store with one-on-one support instruction on reading food ingredient labels.
  • Alternative recipes for preparing favorite meals without corn sweetener ingredients. For example, corn syrup was eliminated as an ingredient in making Indian fry bread.

Participant blood sample analyses

Pre- and post-intervention blood samples were analyzed successfully to determine fasting glucose and insulin levels using standard methods at the local clinical laboratory. Insulin and glucose measurements for each participant were entered into the Hepatitis C Society’s online homeostasis model of assessment for insulin resistance (HOMA-IR) calculator [51] for determining the degree of insulin resistance.

The pre-intervention blood testing was performed at the Mayo clinic but because measurements were reported as < 1, 1 or > 1 xg/g mercury (Hg), they were not useful for our purpose which was to measure lower detectable Hg exposures.  Post-intervention Hg samples were sent to Duquesne University for analysis using a method developed for and used by the Centers for Disease Control and Prevention (CDC) [52]. These samples were analyzed for total Hg based on microwave enhanced sample digestion (EPA Method 3052) with direct isotope dilution mass spectrometry, D-IDMS (EPA Method 6800) using Agilent 7700 inductively coupled plasma mass spectrometry (ICP-MS). In addition, two of the post intervention blood samples  were analyzed for Hg2+ and CH3Hg+ based on microwave enhanced extraction (EPA Method 3200) with speciated isotope dilution mass spectrometry, SIDMS (EPA Method 6800) using Agilent 7890 gas chromatograph (GC) connected to Agilent 7700 ICP-MS. The method was capable of detecting Hg species down to ng/g levels. Each of these EPA numbered methods were developed by the same research laboratory at Duquesne University and validated and published by the Environmental Protection Agency (EPA) under Resource Conservation and Recovery Act (RCRA) statues [53].

Statistical analysis of pilot study data

Statistical analysis was performed using Excel software. Results are expressed as mean difference and standard deviation (SD). A p value of <0.01 is considered significant. A one tailed t-test analysis was conducted to compare the mean of the difference pre and post within each intervention group. In the comparison of the post intervention mercury levels, the online SISA analytical tool [54] was used to conduct the t-test of the means for the two unequal samples (n=5 and n=4). The diet score data in Table 2 was tallied using a scoring method of providing one point for each item if the participant in the MAC intervention course reported a diet habit consistent with the instruction. For example, in scoring each question on highly processed food consumption, participants reporting less consumption were awarded one point. The higher score in this category indicates less consumption of highly processed foods. Conversely, the higher score in the whole foods (minimal processing) and/or organic products category indicates more consumption of foods less likely to contain invasive toxic substances introduced through processing.

Table 2. Food frequency survey diet scores for MAC participants

a score = 1 if “never, rarely (once or twice a month)” b score = 1 if  “once a week, rarely (once or twice a month)” c score = 1 if “once a week, several times a week, every day (1-2 servings)” d score = 1 if “never, rarely, once a week” e score =1 if “several times a week, pretty much every day (1-2 servings), several times a day (3 or more servings)” f score = 1 if “rarely (once or twice a month), once a week, several times a week, pretty much every day (1-2 servings)”

Category

Pre- test

Post- test

t-test

Questions

n=5

n=5

Corn sweeteners , refined or high sugar

How often do you drink a sugar sweetened beverage (do not include diet drinks)?  a

2

3

p>0.05

How often do you drink an energy drink? a

4

4

During the past month, how many times did you eat canned fruit (applesauce, apricot halves, mixed fruit, pears, cling peaches)? a

5

5

During the past month, how many times did you drink 100% fruit juice (apple, orange, grape, cranberry, other)? c

1

1

During the past month, how many times did you eat “sweet snacks” such as candy, cookies, ice cream, popsicle, other sugar sweetened treat (do not include diet)? d

1

5

Total score for category

13

18

Mean

2.6

3.6

SD

1.817

1.673

Fish

p >0.05

How many times in the past month have you eaten freshly caught fish? b

2

1

During the past month, how many times did you eat canned tuna? b

3

5

During the past month, how many times did you eat canned salmon? c

0

0

Total score for category

5

6

Mean

1.66

2

SD

1.528

2.646

Whole foods (minimal processing) and/or organic products

During the past month, how many times did you eat fresh or frozen fruit (bananas, oranges, apples, strawberries, etc….)? e

3

4

During the past month, how many times did you eat fresh vegetables (spinach, lettuce, tomato, carrot, green salad, etc…)? e

2

5

Whole foods (minimal processing) and/or organic products

p<0.01

During the past month, how many times did you eat frozen vegetables (corn, broccoli, peas, green beans, etc….)? e

2

3

During the past month, how many times did you eat poultry (chicken or turkey)? e

3

4

How often in the past month, did you eat red meat (hamburger, pork, ham, or sausage)? d (moderation)

2

2

During the past month, how many times did you eat “brown rice”? f

3

4

During the past month, how many times did you eat oats (oat meal)? f

5

4

During the past month, how many times did you eat canned vegetables (green beans, carrots, corn, peas, spinach, sweet potatoes, diced tomato, mixed vegetables)?  b (moderation)

1

2

How many times in the past month did you eat foods prepared with organic flour? f

0

3

How many times in the last month did you eat organic vegetables or fruit (fresh or frozen)? f

4

4

How many times in the past month did you eat organic processed foods (crackers, bread, cereal, canned vegetables, salad dressing, etc….)? f

2

5

Total score for category

27

42

Mean

2.454

3.818

SD

1.368

0.874

Highly processed foods

p<0.01

During the past month, how many times did you eat canned meals (soup, re-fried beans, chili with and without beans, beef stew, etc..)? d

5

5

During the past month, how many times did you eat processed cheese (American)? a

1

3

During the past month, how many times did you eat processed meat (lunch meat, hotdogs, bacon, beef jerky, etc…)? d

3

4

During the past month, how many times did you eat ready-to-eat cereal (corn flakes, rice crisp, corn squares, oat circles, etc….)? d

5

5

During the past month, how many times did you eat foods fried in vegetable oil, lard, or butter, such as potato chips, french fries, fry bread, doughnuts, hash browns, fried eggs, etc…? d

1

3

During the past month, how many times did you eat “salty” snacks such as potato chips, pretzels, corn chips, pop-corn, etc…? d

3

5

During the past month, how many times did you eat grain products made of wheat such as macaroni, bread, hamburger buns, hotdog buns, or spaghetti? d

2

3

During the past month, how many times did you eat “white” rice? d

4

5

During the past month, how many times did you eat meals prepared in restaurants, fast food places, pizza parlors, or from vending machines? d

3

4

Total score for category

27

38

Mean

3

4.222

SD

1.414

0.972

Results of NHANES dataset analyses

In the NHANES 1999-2012 dataset, n=16,232, we found a significant, direct relationship between blood inorganic mercury (I-Hg) and fasting blood glucose (Table 3). This was true of both the continuous variable for blood inorganic mercury (I-Hg concentration, p<0.001) and the binary variable, I-Hg detect (OR 1.03, p=0.006). After adjustment for age, race, and gender, I-Hg concentration remains significantly associated (p<0.001) with blood glucose levels in the survey-weighted population. Figure 1 shows these associations that were significant in the raw and survey weighted populations and after adjusting for age, race, and gender. Although blood organic mercury was associated with glucose levels (p<0.001), this correlation fell out and became not significant when adjusted for age, race and gender (p=0.097). However, blood organic mercury was inversely associated with insulin levels even after adjusting for age, race, and gender (p<0.001). Our results indicate that blood glucose and inorganic mercury (I-Hg) share an association that is unique to this mercury speciation. The inverse association between blood organic mercury with insulin suggests a complex relationship between the levels of different mercury forms, glucose and insulin.

Figure 1: Inorganic Hg (I-Hg) detection is directly associated with fasting glucose in the NHANES 1999-2012 dataset.

Legend: A. Display of logistic regression illustrates that the probability of I-Hg detection in blood (above 0.4ug/L) is directly associated with fasting glucose levels in the raw NHANES population (Odds Ratio 1.02114, p=0.034, N=16,232). B. Same as A except this is a multivariate analysis that adjusts for race and sex (Odds Ratio 1.02886, p=0.004, N=16,232). C. Same as A except this is the survey weighted population (Odds Ratio 1.03395, p=0.006, N=16,232, SP=4.137e+08). D. Same as B except this is a survey weighted population (Odds Ratio 1.04260, p=0.001, N=16,232, SP=4.137e+08).

Table 3. Mercury (Hg) as a function of glucose (mM) in NHANES 1999-2012 Survey Weighted Population

Population

Hg Species

Outcome

Regression

Correlation

Confidence

P-value

N=16,232, SP=4.137e+08

Blood I-Hga

Glucose (mM)

Linear

 β =0.00260

CI (0.00234-0.00286)

p<0.001

N=16,232, SP= 4.137e+08 Adj. for age, gender, race

Blood I-Hga

Glucose (mM)

Linear

 β =0.00150

CI (0.00123-0.00178)

p<0.001

MA (vs CAU)

Linear

β =0.00184

CI (0.00078-0.00289)

p=0.001

Black (vs CAU)

Linear

β =0.00394

CI (0.00312-0.00475)

p<0.001

N=16,232, SP=4.137e+08

Blood I-Hgb

Glucose (mM)

Logistic

OR=1.03395

CI (1.00980-1.05867)

p=0.006

N=16,232, SP=4.137e+08
Adj. for race and gender

Blood I-Hgb

Glucose (mM)

Logistic

OR=1.04260

CI (1.01674-1.06911)

p=0.001

MA (vs CAU)

Logistic

OR=0.70323

CI (0.58613-0.84372)

p<0.001

N=16,230, SP=4.137e+08

Blood Organic Hga

Glucose (mM)

Linear

β = 0.05923

CI (0.03654- 0.08193)

p<0.001

N=16,230, SP=4.137e+08 Adj. for age, gender, race

Blood Organic Hga

Glucose (mM)

Linear

β = 0.01952

CI (-0.00364- 0.04268)

p=0.097

N=16,230, SP=4.137e+08

Blood Organic Hga

Insulin (pM)

Linear

β = - 0.00121

CI (-0.00165-  -0.00076)

p<0.001

N=16,230, SP=4.137e+08 Adj. for age, gender, race

Blood Organic Hga

Insulin (pM)

Linear

β = - 0.00120

CI (-0.00161- -0.00078)

p<0.001

N = Raw Population, SP = Survey Population, Adj = adjusted, I-Hg = Inorganic Mercury, ug/L = micrograms/liter, mM = millimolar, β = coefficient, a refers to concentration of mercury(ug/L), b refers to detection above 0.4 ug/L blood I-Hg, CI = Confidence Interval , MA = Mexican American, CAU = Caucasian, OR = Odds Ratio,  pM = picomolar

Results of Health Education Intervention