Dietary Patterns and Nutritional Status in Relation to Consumption of Chickpeas and Hummus in the U.S. Population

Chickpeas, a commonly consumed legume, are the main ingredient in traditional hummus. U.S. dietary guidelines recommend consuming 1–1.5 cups of legumes per week. This study aimed to evaluate temporal changes in hummus and chickpea consumption and describe diet and biomarkers of health in U.S. consumers versus non-consumers. National Health and Nutrition Examination Survey (2005–2016) data were used. Dietary intake was collected using two 24-h recalls; age, gender, and poverty-to-income ratio were adjusted in statistical analyses. The proportion of the population who consumed chickpeas or hummus increased significantly over time. Hummus and chickpea consumers were more likely to obtain recommended legume intake. Consumers had significantly increased intakes of fruits, vegetables, and whole grains and decreased added sugars intakes versus non-consumers. Consumers also had lower concentrations of urinary iodine and higher concentrations of serum 4-pyridoxic acid, total vitamin B6, and red blood cell folate. Adults who consumed chickpeas and hummus were 48% and 62% less likely to have metabolic syndrome, respectively. Consuming chickpeas or hummus may be a practical means of improving diet quality and nutritional status. Future work should evaluate whether chronic disease incidence is reduced by chickpea and hummus consumption through better nutrition or lower metabolic syndrome incidence.


Introduction
Dietary pulses, the edible seeds of legumes (e.g., beans, lentils, chickpeas, and peas), are sustainable plant-protein sources that are also high in fiber and various micronutrients [1][2][3]. The Dietary Guidelines for Americans (DGA) first recommended the inclusion of legumes in the 2005 iteration [4]. The 2015 DGA now recommends that Americans consume 1-1.5 cups of legumes per week [5]. Despite these recommendations, toddlers aged 1-3 years are the only group in the United States that consumes adequate amounts of legumes [5]. This is relevant given that legume and pulse consumption has been associated with a wide array of health benefits, particularly reduced cardiometabolic risk when consumed as part of a healthy overall dietary pattern [6][7][8][9][10][11][12]. The inverse relationship between legumes and risk of coronary heart disease is apparent at doses ≥4 servings per week among prospective cohort studies [11,12]. Further, doses of about 1 serving per day have shown meaningful reductions in cardiometabolic risk factors such as body weight, blood pressure, and hemoglobin A1C [6][7][8]12]. Chickpea (Cicer arietinum L.), also commonly known as garbanzo beans, is a commonly consumed pulse that has grown popular in Western culture due to its nutty

Materials and Methods
Data from the National Health and Nutrition Examination Survey (NHANES) (2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016) were used for analysis. NHANES is a nationally representative survey of the non-institutionalized U.S. population that is conducted on an ongoing basis, with data releases provided in 2-year cycles [20,21]. Data collection included interviews conducted in the home, in the Mobile Examination Center (MEC), and over the phone and in-person measurements of body measures and biospecimens. Publicly available data were obtained from the Centers for Disease Control and Prevention National Center for Health Statistics [20,21].

Demographic and Socioeconomic Characteristics
Age and gender were self-reported variables in NHANES. Age was reported to the nearest year, and all individuals aged ≥85 were classified as age 85 years to ensure confidentiality in the public release data. Gender categories included in NHANES were male and female. There were no missing data for age or gender in NHANES. Individuals were excluded if they were younger than age 2 years. Family poverty-to-income ratio (PIR) was calculated based on reported income, and individuals with a PIR > 5 were classified as having a PIR of 5 to ensure confidentiality. In this analysis, PIR was used to estimate socioeconomic status. Individuals were excluded if PIR was missing.

Dietary Intake
Dietary intake in NHANES participants was ascertained through two 24-h recalls. The first 24-h recall (referred to as day 1 intake) was conducted in the MEC, and the second 24-h recall (day 2 intake) was conducted via telephone 3-10 days after the first recall [22]. Dietary intake data were provided as individual foods files and total nutrient intake files. Individual foods files were used to identify and estimate total hummus and chickpea consumption and identify meals in which foods were consumed. Individuals were excluded if there were missing data for either dietary recall or if the average total kilocalorie consumption across the two 24-h recalls was <500 or >5000 kcal.
Individual foods were identified as hummus or chickpeas based on the U.S. Department of Agriculture (USDA) eight-digit food code (Table 1). This information is used to classify individual foods reported by the individual as either being or containing hummus or chickpeas, and from that information, the classification of being a consumer and how much can be made. Each food reported by an individual was classified as hummus or chickpea. The total amount of hummus and chickpea foods consumed at each meal was calculated for each person in grams. There were 20 possible meal types in NHANES that include options in Spanish and extended consumption, and meals were classified into 5 groups: breakfast (breakfast, desayuno), lunch (lunch, brunch, almuerzo), dinner (dinner, supper, cena), snack (snack, entre comida, botana, bocadillo, tentempié), and other (drink, infant feeding, extended consumption, comida, merienda, bebida, other). Individuals were classified as consuming hummus or chickpeas on day 1 and day 2. Each individual was also classified as having consumed hummus or chickpeas on either day. Chickpeas, canned, drained, fat not added in cooking X 41302100 Chickpeas, canned, drained, low sodium, NS as to fat added in cooking X 41302110 Chickpeas, canned, drained, low sodium, fat added in cooking X 41302120 Chickpeas, canned, drained, low sodium, fat not added in cooking X 41310150 Stewed chickpeas, Puerto Rican style X 41310160 Stewed chickpeas, with potatoes, Puerto Rican style X 41310200 Chickpeas stewed with pig's feet, Puerto Rican style X 41310210 Stewed chickpeas with Spanish sausages, Puerto Rican style X 41310220 Fried chickpeas with bacon, Puerto Rican style X 41602020 Garbanzo bean or chickpea soup, home recipe, canned or ready-to-serve X 75302029 Beans, string, green, with chickpeas, cooked, NS as to fat added in cooking X 75302030 Beans, string, green, with chickpeas, cooked, fat not added in cooking X 75302031 Beans, string, green, with chickpeas, cooked, fat added in cooking X 1 NS: not specified; USDA, U.S. Department of Agriculture.
The USDA provides food pattern equivalents for NHANES, and this information was merged with the other NHANES files [23]. Food pattern equivalents for vegetables (cup equivalents), fruits (cup equivalents), whole grains (ounce equivalents), and added sugars (in grams) were used to describe dietary characteristics to compare across hummus and chickpea consumers. Additional food pattern equivalents were used to calculate total legume consumption, and individuals were classified as meeting legume recommendations based on their average kilocalorie intake over the 2 days of recalls as follows: ≥0.5 cup-equivalent/week for <1500 kcal, ≥1 cup-equivalent/week for ≥1500 to <1700 kcal, and ≥1.5 cup-equivalent/week for ≥1700 kcal.

Anthropometry and Blood Pressure
Trained technicians took NHANES participants' physical measurements according to an established set of protocols [20]. All individuals were eligible for the body measure components of the NHANES in-person visit. Measured weight (in kilograms) and waist circumference (in centimeters) and calculated body mass index (BMI; in kilograms per square meter) based on measured weight and height were used in this analysis. Three consecutive blood pressure measurements were taken.
Blood pressure was measured in NHANES participants aged >8 years. The average of available measurements for systolic and diastolic blood pressure (in millimeters of mercury) was used in this analysis (e.g., average of 3 measurements if all 3 were available; average of 2 measurements if only 2 were available).

Urine and Blood Biospecimens
NHANES participants provided urine and blood samples during the in-person visit to the MEC. Participants were not required to be fasted, so a subset of participants who happened to be fasted were eligible for fasting measurements. Measurements from urine and blood biospecimens used in the analysis for the evaluation of dietary biomarkers included urinary iodine and serum selenium, copper, zinc, vitamin D (D 2 + D 3 ), vitamin B 12 , vitamin B 6 (4-pyridoxic acid, pyridoxal 5 -phosphate, and serum B 6 ), and red blood cell folate.
Fasting blood biospecimens were also evaluated for high-density lipoprotein (HDL) cholesterol, triglycerides, and glucose according to established protocols [20], and these values were used in the calculation of metabolic syndrome (MetS) classification.

Metabolic Syndrome
MetS was calculated for adults only (age ≥ 20 years) due to a lack of collection of all the component biomarkers in younger populations, per NHANES protocols. Individuals were classified as having MetS based on having 3 or more of the following 5 National Cholesterol Education Program's Adult Treatment Program III (ATPIII) criteria: abdominal obesity (waist circumference, >88 cm for women and >102 cm for men), high triglycerides (fasting triglycerides, >150 mg/dL), low HDL (<50 mg/dL for women, <40 mg/dL for men), high blood pressure (systolic, ≥135 mm Hg; diastolic, ≥85 mm Hg), and high fasting blood glucose (≥110 mg/dL) [24].

Statistical Analysis
There were 60,936 individuals in the NHANES 2005-2016 data releases. The following exclusions were applied for the main analyses (in order of application) to include 37,552 individuals: age ≤ 2 years (n = 6280), no day 2 dietary recall (n = 12,524), missing PIR data (n = 3042), average intake < 500 kcal (n = 113), average intake > 5000 kcal (n = 161), missing anthropometry data (n = 1261), and missing dietary intake data for day 1 (n = 3). Smaller numbers of individuals were included for biomarker measurements based on the years of the data available and the availability of measurements for individuals. The data release years and sample sizes for these analyses were as follows: urinary iodine  [20]. Where possible, data from all data releases were used; however, for some of the urinary and blood measurements, these variables were only available for a smaller number of data releases as previously indicated, and the population weights were divided by the number of cycles utilized for the analysis. For descriptive statistics, frequencies and population-weighted percentages were calculated for dichotomous or categorical outcomes. Means and population-weighted standard errors were calculated for numerical outcomes. Arithmetic means were used for variables that were reasonably normally distributed upon visual inspection of the histogram. Geometric means were used for variables that had a skewed distribution. Multivariable logistic or linear regression analyses were used to model associations for dichotomous and numerical dependent variables, respectively, with age, gender, and PIR included as adjustment variables. Analyses were conducted using Stata software (version 15; StataCorp LLC, College Station, TX, USA).

Temporal Changes in Hummus and Chickpea Consumption in the U.S. Population
The percentage of individuals who reported hummus or chickpea consumption has increased in the population over time (Figure 1). In the population as a whole, there were statistically significant increasing trends in the percentage of consumers of hummus (p-trend < 0.001) and chickpeas (p-trend < 0.001) when adjusted for age, gender, and PIR. Females had higher hummus and chickpea consumption than males, but this was less apparent in more recent years; both groups had significant increases over time in the percentage of population who were consumers of hummus (p-trend = 0.002 for males and 0.001 for females) and chickpeas (p-trend = 0.007 for males and 0.001 for females), adjusted for age and PIR. With regard to age, children had the lowest percentage of hummus and chickpea consumption. All age groups exhibited increases in the percentage of hummus and chickpea consumers over time (p-trend < 0.05 for all groups). Higher-income individuals had the highest percentages of hummus and chickpea consumers in the population. After adjustment for gender and age, middle-income and higher-income groups had increased percentages of hummus and chickpea consumers over time (p-trend < 0.05 all groups), and lower-income groups had an increase in the percentage of hummus consumers (p-trend < 0.05) but not chickpea consumers.

Dietary Patterns of Hummus and Chickpea Consumers
Across
Of individuals who reported hummus consumption, the majority reported consuming it once per day (91% for day 1, 89% for day 2) ( Table 2). No one reported consuming hummus >4 times per day. Among hummus consumers for each of the days, the most frequently reported meals for consumption were lunch, dinner, and snacks. A small percentage of individuals indicated consumption of hummus at breakfast and other times, and the mean amount consumed at breakfast was lower than other eating occasions.  1 Values are presented as percentages or means (SE). Other meal occasions are not presented due to low frequency (<1% of population). NHANES, National Health and Nutrition Examination Survey; 2 Percentages sum to >100 because some individuals consumed hummus or chickpeas more than once per day.
Of the individuals who reported chickpea consumption, the majority reported consuming it once or twice per day (93% on day 1, 92% on day 2). The highest frequency of chickpea consumption was eight times per day. Chickpeas were most often consumed at lunch, dinner, or snacks. Small percentages of individuals reported eating chickpeas at breakfast and other occasions.
A higher percentage of individuals who consumed hummus on either day met the recommended intakes of legumes (69.7%) versus individuals who did not consume hummus (23.9%). This difference represented an eight-times higher likelihood of meeting legume consumption recommendations among hummus consumers, after adjustment for age, gender, and PIR (OR = 7.8 [95% CI, 5.6, 10.9]). A higher percentage of individuals who consumed chickpeas on either day met recommended intakes of legumes (72.6%) versus individuals who did not consume chickpeas (23.5%). This difference represented a nine-times higher likelihood of meeting legume consumption recommendations in chickpea consumers, after adjustment for age, gender, and PIR (OR = 9.2 (95% CI, 7.0, 12.0)). Hummus and chickpea consumers had higher intakes of fruits, total and dark green vegetables, total protein foods, whole grains, and lower intakes of meat and added sugars compared to non-consumers, adjusted for age, gender, and PIR (Table 3). No differences across groups were observed for refined grain or dairy consumption. 1 Values are presented as means (SE) of two-day intake. NHANES, National Health and Nutrition Examination Survey; 2 p-value for the difference in means across consumers and non-consumers, adjusted for age, gender, and poverty-to-income ratio.

Health Outcomes in Hummus and Chickpea Consumers Versus Non-Consumers
There were some observable differences in dietary biomarkers across hummus and chickpea non-consumers and consumers (Table 4). Hummus and chickpea consumers had lower concentrations of urinary iodine and higher concentrations of serum 4-pyridoxic acid, total vitamin B 6 , and red blood cell folate. There were no significant differences in serum concentrations of pyridoxal 5 -phosphate, vitamin B 12 , vitamin D (25OHD 2 , 25OHD 3 , and 25OHD 2 + 25OHD 3 ), copper, selenium, or zinc. There were no significant differences in body weight across hummus and chickpea consumers and non-consumers (Table 5). However, hummus and chickpea consumers had significantly lower BMI and waist circumference than non-consumers, after adjustment for age, gender, and PIR. MetS could only be classified in adults. A lower percentage of hummus consumers (either day) had MetS (n = 13, 7.0%), compared to hummus non-consumers (n = 2742, 20.2%). After adjustment for age, gender, and PIR, hummus consumers were 62% less likely to be classified as having MetS (OR = 0.38 [95% CI, 0.16, 0.88]). A lower percentage of chickpea consumers (either day) had MetS (n = 33, 9.8%) compared to chickpea non-consumers (n = 2721, 20.2%). After adjustment for age, gender, and PIR, chickpea consumers were 48% less likely to be classified as having MetS (OR = 0.52 [95% CI, 0.31, 0.89]).

Discussion
In the current study, we found that hummus and chickpea consumption has increased significantly since the 2005-2006 NHANES; however, overall consumption in the population is low. The finding that hummus, but not chickpea, consumption has increased since the 2005-2006 NHANES among lower-income individuals is not surprising due to the demonstrated limited access to similar items such as fresh produce [25,26]. On the other hand, the minimal food processing techniques involved with extending the shelf life of hummus [27,28] make it a reasonable strategy to increase intake of legumes among lower-income populations. The further observation that consumption of hummus and chickpeas is independently associated with increased intakes of fruits, vegetables, total protein foods, and whole grains (and decreased intake of added sugars) supports consumption to be an effective strategy to increase intakes of other healthful foods in the diet. According to the Produce for Better Health Foundation, recent fruit and vegetable consumption declines are tied to two common behaviors: a decline in dinner side dishes for vegetables, and a reduced consumption of fruit juice at breakfast [29]. Hummus and chickpeas were shown to be mostly consumed at lunch and dinner, as well as during snacking occasions. More than 90% of Americans report two to three snacking occasions throughout the day; snacking provides nearly 23% of total daily energy intakes for most Americans and as much as 35% of total added sugars among children [30]. It is likely that hummus can act as a "carrier" food to help increase diet quality and the intake of more healthful foods such as vegetables and whole grains during snacking and other eating occasions. A recent study in healthy adults supports this observation; an afternoon hummus snack was associated with improved diet quality and a ≈20% decrease in subsequent snacking of desserts [19].
Based on nutrient biomarker analyses, consumers of hummus were more likely to have higher levels of serum 4-pyridoxic acid, total vitamin B 6 , and red blood cell folate. We recently showed that serum vitamin B 6 levels decrease with age among middle-aged to older adults [31], indicating a potential need for future observational data that extend beyond the capabilities of NHANES and future public emphasis on foods (or dietary supplements) that contain significant amounts of this essential nutrient. The finding that hummus and chickpea consumers have a lower incidence of MetS and decreased BMI and waist circumference is likely somewhat due to its contribution to intakes of several essential macro-and micronutrients associated with increased diet quality [14]. As supported by the other observations, hummus and chickpea consumption is also likely a marker of healthier diet and lifestyle habits.
A major strength of this study is the use of NHANES, which has a large sample size and is nationally representative. Limitations of NHANES include the sole reliance on self-reported dietary data. A previous validity analysis of the USDA Automated Multiple-Pass Method used for collecting 24-h dietary recalls in the What We Eat in America NHANES component found that overall, obese individuals under-report total daily energy intakes by about 11% compared to <3% among normal-weight individuals [32]. While two days of dietary intake may not capture infrequent hummus or chickpea consumers, the results may more reasonably represent usual or frequent hummus or chickpea consumers. NHANES is a series of cross-sectional surveys and cannot be used to calculate incidence of disease or evaluate changes in individuals, and it is not possible to establish a temporal sequence of associations. However, these results provide an informative foundation about potential roles of hummus and chickpea consumption with cardiometabolic diseases. Future studies incorporating longitudinal designs can be used to evaluate whether chronic disease incidence is reduced by chickpea and hummus consumption through better nutrition or lower MetS incidence. Additionally, while NHANES is designed to be representative of the US population, selection bias can occur based on the individual participation in studies, and it commonly held that healthy individuals participate in population-based health research studies, and there may be some limitations on external validity. However, NHANES is designed to be representative of the US population and is used by the Centers for Disease Control and Prevention to assess health trends in the US population as it provides a reasonable representation of the US population.

Conclusions
Consumption of hummus and chickpeas in the United States has increased since the 2005-2006 NHANES. Cross-sectional data indicate that hummus and chickpea consumption may lead to more optimal intakes of fruits, total and dark green vegetables, total protein foods, and whole grains, and lower intakes of meat and added sugars compared to non-consumers, adjusted for age, gender, and PIR. No differences across groups were observed for refined grain or dairy consumption. Consumption in this dataset was also associated with an improved status of serum vitamin B 6 and red blood cell folate and a decreased incidence of MetS.