Dietary Intake, Behaviors and Psychosocial Factors Among Women from Food-Secure and Food-Insecure Households in the United States

Abstract

Objective

Determine whether macro- and micro-nutrient intake, energy intake, diet quality, adherence to recommended dietary intake, and psychosocial and behavioral factors are associated with household food security.

Design

Baseline data from in-person interviews and telephone-based, 24-hour dietary recall from women recruited to a diet and physical activity controlled trial.

Setting

Neighborhoods encompassing 18 urban census tracts in South Carolina.

Participants

Participants (n=202) were predominantly African American (87%), overweight or obese women aged 25 to 51 years with mean body mass index of 40.6±8.7.

Main Outcome Measures

Macro- and micro-nutrient intake, energy intake, diet quality, adherence to recommended dietary intake (via multi-pass, 24-h recall); diet-related self-efficacy and social support, healthy/lowfat and emotional eating behaviors, and depressive symptoms.

Results

Women in food-secure (FS) and food insecure (FI) households were not different on health and sociodemographic characteristics. Women in FI households had lower self-efficacy and healthy/low-fat eating behaviors, and higher emotional eating and depressive symptoms compared with women in FS households. The groups did not differ on social support. Significant dietary differences were few (FS>FI on protein and lean meat; FS<FI on carbohydrate intake). For 29 of 35 (74%) dietary intake recommendations, less than 75% of women in both groups met each recommendation.

Conclusions

While food security status was associated with diet-related psychosocial and behavioral factors, it was associated with few aspects of dietary intake. Dietary intake overall was poor. Participants were not meeting guidelines for a diet supportive of general health or weight loss and management, regardless of food security status.

Introduction

Disparities in obesity, chronic disease risk, and diet quality are well-documented between African American and White women in the United States. Non-Hispanic Black women (58.6%) are significantly more likely to be overweight or obese than non-Hispanic White (33.4%) and Hispanic (40.7%) women1 and to have diets of lower nutritional quality.2,3

Observational studies suggest that food insecurity is associated with obesity,4-8 chronic diseases and pregnancy complications.9-11 The food insecurity-obesity relationship has been found most consistently among women;4,5,8 the evidence among men and children is mixed.12 The association has not been found consistently across racial and ethnic groups. Battacharya and colleagues7 found an association between food insecurity and obesity among Hispanics and Whites but not among African Americans in a national sample.

Across diverse studies, findings have been inconsistent regarding the association between food security status and dietary factors. In total, these studies suggest that food insecurity is predictive of a poor diet;7,13-22 however, few examined this association among or within racial and ethnic groups. National data showed that poverty-level income was associated with lower diet quality among both White and Black, but not Hispanic, respondents, while food insecurity predicted lower diet quality only among Whites,7 suggesting that income and food security are not synonymous in their associations with dietary outcomes, nor consistent across racial and ethnic groups. Further, while observational studies indicate that neighborhood poverty is associated with lower availability of and access to healthy foods, there is no clear evidence of a causal association between food availability/convenience and health or diet-related outcomes across racial or ethnic subgroups.23 The specific interrelationships among food insecurity, poverty, race/ethnicity, community environment and obesity among and within these subgroups remain unclear.

The purpose of our study was to determine whether dietary intake and quality, energy intake, adherence to recommendations, and psychosocial and behavioral factors differed by household food security status (food secure vs insecure) among mainly African American women who were overweight or obese and recruited from census tracts of high poverty in central South Carolina. This sample allowed us to test for an association between food security status and dietary intake within a relatively homogeneous sample of participants. While the study was primarily descriptive, we hypothesized that women in food-secure households would have a more positive diet-related psychosocial status, better macro- and micronutrient intake and diet quality, and better adherence to recommended dietary intake levels than women in food-insecure households.

Methods

Baseline data were from a randomized controlled trial,24 a behavioral intervention to reduce body weight, increase physical activity, and improve dietary intake. Women aged 25-50 years were recruited from 18 census tracts in Columbia, South Carolina, within a Standard Metropolitan Statistical Area of more than 500,000. Of 61 tracts, these tracts had ≥ 25% (25%-62%) of residents with below-poverty income. Eligibility included body mass index ≥25 kg/m²and waist circumference ≥88 cm (detailed inclusion criteria are in print).24

A community advisory board of women leaders from the neighborhoods assisted with recruitment. Staff persons screened potential participants by phone. Initially eligible women attended in-person measurement. Staff persons collected baseline data from three cohorts during November through mid-December, 2008 and October through mid-December, 2009 and 2010. For ethical reasons, food-insecure women received food assistance referrals.

Main Measures

Sociodemographic and Health Variables

Participants self-reported their age, number of dependent children at home, race, Hispanic ethnicity, education, employment, health insurance, 10 health conditions (yes/no) and marital status. The Actigraph accelerometer, GT1M model, (ActiGraph, LLC, Fort Walton Beach, FL) assessed seven days of physical activity.

Body Mass Index

Staff obtained height to the nearest quarter inch and weight to the nearest tenth kilogram. Body mass index was calculated as weight in kg/height in m^2.

Waist Circumference

Staff used the iliac crests as landmarks for waist circumference to the nearest tenth centimeter.25

Food Security Status

The six-item short form of the 12-month Food Security Scale26 measured financially based food security. The short form classifies 97.7% households correctly and underestimates overall food insecurity by .3%.26 Scoring categorized households as food secure or insecure.

Depressive Symptoms

The validated short form of the Center for Epidemiological Studies Depression Scale (CESD-10)27 assessed depressive symptoms. A cut-point of 10 has shown sensitivity of 77% and specificity of 79% relative to a structured clinical interview.28 Cronbach’s alpha in this sample was .83.

Dietary Intake

Registered dieticians trained in the University of Minnesota’s Nutrient Data System for Research (NDSR) protocols conducted three 24-hour dietary recall interviews by telephone (two week days and one weekend day). The multi-pass, 24-hour dietary recall methodology has established validity and reliability.29,30 Participants received a 20-minute training in portion size estimation using a Food Portion Visual,31 with the addition of food models, dishes and utensils, then completed their first 24-hour dietary recall. The remaining two recalls occurred within 15 days. Average time between the first and third interviews was 9.5 days. Dietary data were collected and analyzed using Nutrition Data System for Research software version 5.0_35, (2005), Nutrition Coordinating Center, University of Minnesota, Minneapolis, MN. Serving sizes and referent intake levels were defined by the NDSR manual©32 and 2005 Dietary Guidelines for Americans,33 Institute of Medicine34,35 and American Heart Association,36 as these guidelines were in effect when data collection began.

Self-Efficacy for Healthy Eating

An eight-item self-efficacy scale for eating a low-fat diet37 was modified to assess self-efficacy for “making healthy food choices, such as choosing fruits or vegetables, choosing lower fat foods, and watching how much you eat” when faced with common barriers. The original scale demonstrated construct, convergent validity, and Cronbach’s alpha > .82.37 Cronbach’s alpha in this sample was .75.

Social Support for Healthy Eating

In a separate sample of middle-aged SC women (60% African American), we created and validated a measure of social support for healthy eating. Respondents rated to what extent “family members, friends, co-workers, or anyone else close to you” engaged in 11 supportive behaviors. In the validation sample, the scale had good construct validity, with factor loadings from .65 to .82 and Cronbach’s alpha of .93. Cronbach’s alpha in this sample was .92.

Healthy/Low-Fat Eating

We modified eight items from the Eating Behavior Patterns Questionnaire to assess healthy and low-fat eating behaviors.38 The original subscale had construct validity and internal consistency reliability (Cronbach’s alpha = .84) among African American women.38 Cronbach’s alpha in this sample was .70.

Emotional Eating

The tendency to eat in response to negative emotions was assessed with four items (eating when upset, for comfort, when not hungry, eating until package of food was finished) from the emotional eating subscale of the Eating Behavior Patterns Questionnaire.38 This subscale had acceptable internal consistency reliability (α = .77) and construct validity in African American women.38 In this sample, Cronbach’s alpha was .66. Higher scores indicated less emotional eating.

Diet Quality

The Alternative Healthy Eating Index (AHEI) has been strongly associated with health outcomes.39,40 The original AHEI consists of nine components: vegetables, fruit, non-meat proteins, ratio of white to red meat, cereal/grain-based fiber intake, trans-fat, ratio of polyunsaturated to saturated fat, duration of multivitamin use, and alcohol use. We did not have data on multi-vitamin use, so the possible range of scores was 0 to 80, with a higher score indicating a higher-quality diet.

Statistics

The Statistical Analysis System 9.2 (Cary, NC) computed means, standard deviations, frequencies, and percentages. T-tests and chi-squared tests (or Fisher’s exact test for cell sizes <5) were computed to determine comparability of women from food-secure and food-insecure households on socio-demographic, health-related, psychosocial and dietary variables.

Results

Of calls attempted to 746 potential participants who responded to recruitment materials, 657 women were screened for eligibility, 307 met inclusion criteria, and 230 provided informed consent; 26 women were excluded after consent because of medical contraindications and other exclusion criteria. Of the remaining 204 women, two did not complete the dietary recall.

The sample (n=202) was predominantly African American (87.1%) and most were obese (94.5%). The women from food-secure (n=124) and food-insecure (n=78) households were not significantly different on any of the health, anthropometric and sociodemographic characteristics shown in Table 1.

Table 1. Health and sociodemographic characteristics among women from food-secure and food-insecure households.

		Total, n=202		Food Secure, n=124		Food Insecure, n=78		Test statistic
Characteristic		Mean	SD	Mean	SD	Mean	SD	t	P
Age, y		38.2	7.6	38.3	7.7	37.9	7.5	.36	.72
Children <18 y in household		1.0	1.2	1.0	1.3	1.1	1.1	.62	.54
Waist circumference, cm		117.0	17.3	116.4	17.4	117.9	17.1	.62	.53
Weight, kg		109.6	26.0	109.5	26.4	109.9	25.7	.10	.91
Body mass index, weight kg/ height m2		40.6	8.7	40.4	8.6	40.9	8.9	.37	.71
Self-reported medical condition (0-10)a		1.2	1.1	1.2	1.1	1.2	1.2	-.23	.82
		nb	%b	nb	%b	nb	%b	Fisher’s exact	P
Ethnicity	Hispanic	6	3.0	5	4.1	1	1.3		.41
								X2	P
Race	White	16	7.9	11	8.9	5	6.4	7.8	.68
	African American	176	87.1	106	85.5	70	89.7
	> One race	10	5.0	7	5.7	3	3.9
Education	< High school	11	5.5	4	3.2	7	8.9	3.1	.21
	High school/GED	31	15.4	19	15.3	12	15.4
	> High school	160	79.2	101	81.5	59	75.6
Employment	Employed	148	73.3	91	73.4	57	73.1	.14	.93
	Not employed	37	18.3	22	17.7	15	19.2
	Student	17	8.4	11	8.9	6	7.7
Marital status	Not married	86	42.6	47	37.9	39	50.0	4.3	.23
	Married	42	20.8	30	24.2	12	15.4
	Divorced/separated	56	27.7	34	27.4	22	28.2
	Unmarried couple	18	8.9	13	10.5	5	6.4
Health insurance	Yes, has coverage	153	75.7	99	79.8	54	69.2	2.9	.09
Physical activity	< recommendationc	178	90.8	110	91.7	68	89.5	.27	.60

a. Has a doctor told you that you have now or have had in the past (yes, no): heart condition, stroke, cancer, diabetes (“high sugar”), high blood pressure/hypertension, skeletal or muscle injury, arthritis or autoimmune disease, peripheral artery disease, lung condition (asthma, dyspnea, shortness of breath, chronic obstructive pulmonary disease), kidney disease.

b. Total N may not be 202 for every characteristic because of missing data; rounded percentages.

c. ≥5 days of 30 min. moderate-intensity or ≥3 days of 20 min. vigorous-intensity physical activity week, in bouts of ≥10 min

As shown in Table 2, women in food-insecure households had lower self-efficacy for healthy eating, lower healthy/low-fat eating behaviors, and higher emotional eating (ie, lower scores) and depressive symptoms compared with women in food-secure households. Social support was unrelated to food security.

Table 2. Comparison of psychosocial and behavioral factors between women in food-secure and food-insecure households.

	Food-secure			Food-insecure
Characteristic	n	Mean (SD)	Min, Max	n	Mean (SD)	Min, Max	t	P
Depressive symptoms scorea	124	8.3 (5.0)	2.0, 24.0	78	10.9 (6.1)	2.0, 29.0	3.36	.001
Self-efficacy for healthy eatingb	124	19.2 (4.5)	9.0, 29.0	78	17.6 (4.0)	10.0, 29.0	2.55	.01
Social support for healthy eatingc	120	31.1 (10.5)	11.0, 53.0	77	28.6 (10.4)	11.0, 53.0	1.64	.10
Healthy/low-fat eating scored	124	23.5 (5.5)	8.0, 36.0	77	21.9 (6.4)	9.0, 32.0	1.97	.05
Emotional eating scoree	124	11.4 (3.8)	4.0, 20.0	78	10.2 (3.1)	4.0, 17.0	2.45	.02

a. 0-30, higher scores indicate greater depressive symptoms

b. 8-32, higher scores indicate greater self-efficacy for healthy eating

c. 11-55, higher scores indicate greater social support for healthy eating

d. 8-40, higher scores indicate healthier/ lower-fat eating behaviors

e. 4-20, higher scores indicate lower levels of emotional eating

Table 3 shows mean intake for selected nutrients and food groups and the modified AHEI score. Women from food-insecure households had significantly lower intake of protein and lean meat, and significantly higher carbohydrate intake compared with women in food-secure households but showed no other significant differences.

Table 3. Comparison of dietary intake between women in food-secure and food-insecure households.

	Food secure n=124				Food insecure n=78
Dietary intake	Mean	SD		Min, Max	Mean	SD		Min, Max		t	P
Kcals/d	1906		825	645, 7174	1955		656		717, 3747	.45	.65
AHEI, modified	30.8		9.8	10.4, 57.9	28.6		8.8		12.8, 57.9	1.7	.10
% of total kcals/d
Total fat	35.4		5.9	19.4, 55.6	34.5		6.4		18.8, 50.4	1.10	.27
Saturated fat	11.2		2.5	5.4, 22.0	11.2		2.8		5.1, 18.2	.1	.91
Trans fat	2.2		1.1	.3, 6.1	2.1		1.0		.5, 4.7	.35	.73
Protein	16.2		4.3	7.3, 35.8	15.1		3.2		8.55, 24.6	2.02	.05
Carbohydrate	47.7		8.0	28.7, 70.0	50.1		7.8		34.1, 66.8	2.06	.04
Grams/d
Added sugars	87.6		73.4	.8, 665.9	100.6		54.0		3.64, 254.8	1.35	.18
Fiber	13.4		6.4	3.0, 31.0	12.8		6.4		3.1, 37.7	.61	.54
Milligrams/d
Sodium	3251		1343	985, 8003	3105		1084		1437, 5999	.81	.42
Servings/d
Vegetables	2.5		1.4	.3, 7.5	2.3		1.4		0, 5.8	.94	.35
Fruits	1.0		1.0	0, 4.9	.9		1.0		0, 4.2	.30	.76
Fruits & vegetables	3.4		1.7	.3, 12.3	3.2		1.8		0, 10.0	.92	.36
Total grains	5.8		3.0	1.0, 19.8	5.8		2.5		1.6, 13.4	.00	.98
Whole grains	.7		.9	0, 4.5	.7		1.1		0, 4.9	.10	.92
Refined grains	4.8		2.9	0, 19.8	4.9		2.4		.2, 11.1	.17	.87
Dairy	1.1		1.0	0, 5.4	1.2		.9		0, 3.9	.39	.70
Low-fat dairy	.1		.2	0, 1.7	.2		.3		0, 1.2	1.30	.19
Meat	5.2		2.7	0, 13.2	5.3		2.9		.2, 19.7	.17	.86
Lean meat	2.6		2.1	0, 10.3	2.0		1.7		0, 7.5	2.12	.04
Beans	.1		.2	0, 1.3	.1		.2		0, .9	1.50	.13
Meat alternatives	0		.1	0, .6	0		.2		0, 1.2	1.37	.17
Sweetened beverages	2.1		1.8	0, 9.6	2.4		1.7		0, 6.6	1.16	.25
Alcoholic beverages	.1		.3	0, 2.2	.1		.5		0, 4.2	.10	.91

Table 4 shows the results for dietary guidelines. Nearly twice as many women in food-secure households (24.4%) as food-insecure (12.8%) met the recommendation that lean meat comprise ≥75% of total meat intake. No other significant differences were found. Of note is the very small proportion (<25%) of women in both groups who met recommendations that are important for weight control and a healthy diet regarding fiber, fruits and vegetables, whole grains, lean meats, low-fat dairy, fat, added sugars and sweetened drinks. The proportions of all women who met recommendations for calcium, iron, magnesium, potassium, sodium, vitamins D and E, and pantothenic acid were very low (≤25% of women). For 29 of 35 (74%) dietary intake guidelines examined, less than 75% of women in both groups met each recommendation.

Table 4. Proportion of women meeting dietary guidelines by household food security status.

Macronutrient/ food group	Dietary guideline, daily	Total meeting guideline, n=202		Food secure meeting guideline, n=124		Food insecure meeting guideline, n=78		Test statistic
		na	%	n	%	n	%	X2	P
Kilocalories	<2,00033b	123	60.8	81	65.3	42	53.9	2.65	.10
Fatc	20-35% of total kcals33,34	99	49.0	61	49.2	38	48.7	1.01	.60
Saturated fat	<10% of total kcals33	58	28.7	34	27.4	24	30.8	.26	.61
Trans fat	0% of total kcals33	23	11.4	14	11.3	9	11.5	0	.96
Proteinc	10-35% of total kcals33,34	192	95.0	119	96.0	73	93.6	1.75	.42
Carbohydratec	45-65% of total kcals33,34	129	63.9	77	62.1	52	66.7	3.09	.21
Fiber	≥25 grams33,34	12	5.9	7	5.7	5	6.4	.05	.82
Fruits/vegetables	≥5servings33	32	15.8	20	16.1	12	15.4	.02	.89
Whole grains	≥50% of grain intake33	6	2.9	4	3.2	2	2.6	.07	.79
Low-fat diary	≥75% of dairy intake33	8	4.0	3	2.5	5	6.4	1.96	.16
Lean meat	≥75% of meat intake33	40	19.9	30	24.4	10	12.8	4.01	.05
Sweetened drinks	0 servings33	23	11.4	17	13.7	6	7.7	1.72	.19
Added sugars	≤100 kcals33,35	15	6.9	12	9.7	3	3.9	2.37	.12
Alcohol	≤1 serving33	194	96.0	119	96.0	75	96.2	0	.95
Minerals
Calcium	≥1,000 mg.34	27	13.4	18	14.5	9	11.5	.37	.55
Iron	≥18 mg.34	36	17.8	23	18.6	13	16.7	.12	.73
Magnesium	≥320 mg34	25	12.4	13	10.5	12	15.4	1.06	.30
Phosphorus	≥700 mg34	160	79.2	98	79.0	62	79.5	.01	.94
Potassium	≥4,700 mg33,34	1	.04	0	0	1	1.3	1.60	.21
Sodium	<1,500 mg33,34	49	24.3	27	21.8	22	28.2	1.08	.30
Zinc	≥8 mg34	113	55.9	66	53.2	47	60.3	.96	.33
Copper	≥.9 mg34	115	56.9	68	54.8	47	60.3	.57	.45
Selenium	≥55 mcg34	181	89.6	112	90.3	69	88.5	.18	.67
Vitamin A	≥2,333 IU34	114	56.4	71	57.3	43	55.1	.09	.77
Vitamin D	≥15 mcg34	4	19.8	4	3.2	0	.0	2.57	.11
Vitamin E	≥15 mg34	26	12.9	16	12.9	10	12.8	.0	.99
Vitamin K	≥90 mcg34	89	44.1	57	46.0	32	41.0	.47	.49
Vitamin C	≥75 mg34	83	41.1	51	41.1	32	41.0	.0,	.99
Thiamin/B1	≥1.1 mg34	130	64.4	80	64.5	50	64.1	.0	.95
Riboflavin/B2	≥1.1mg34	157	77.7	96	77.4	61	78.2	.02	.90
Niacin/B3	≥14 mg34	169	83.4	101	81.5	68	87.2	1.15	.28
Folate	≥400 mcgf	55	27.2	33	26.6	22	28.2	.06	.81
Vitamin B6	≥1.1mg34	129	63.9	78	62.9	51	65.4	.13	.72
Vitamin B12	≥2.5mcg34	133	65.8	80	64.5	53	68.0	.25	.62
Pantothenic acid	≥5 mg34	51	25.2	32	25.8	19	24.4	.05	.82

a. n=4 women reported no dairy and are omitted from this analysis. n=1 woman reported no meat and is omitted from this analysis.

b. Caloric needs based on Estimated Energy Requirement equations for moderately active female adults between 31-50 years old, per the Institute of Medicine. Overweight and obese women will logically have individualized requirements for weight loss or maintenance.

c. For % of kcals from total fat, protein and carbohydrate, a trichotomy of “below,” “within,” or “above” the recommended intake range was tested. Only 3% of all women exceeded 65% of kcals from carbohydrates, and .5% exceeded 35% of kcals from protein, but 49.5% exceeded 35% of kcals from total fat. All other tests compared “meets” to “does not meet” the recommendation, as defined by cited sources.

Discussion

A strength of this study is the sample’s inclusion of a large proportion of African American and multi-racial women (92%) with high educational attainment, all of whom lived in areas of high poverty in the urban southeastern USA. We obtained 24-hour dietary recall call data from a validated procedure considered the state-of-the-art in community-based studies among participants who are often labelled “hard-to-reach.” Nevertheless, the study has several limitations that must be considered when interpreting the results. There was a lack of temporal congruence between the food security measure (retrospective recall of the past 12 months) and the dietary recall, a common limitation in this literature. We used a short measure of food insecurity and analyzed food security as a dichotomy. Other measures provide multiple categories (eg, secure, marginal, low and very low), but researchers often collapse these. Whether the association of food security to dietary intake approximates a linear trend across categories, or is best represented by some other dichotomy (eg, very low vs all others) or a continuous measure, remains to be investigated.

It is challenging in a single study to measure all potential influences on dietary intake that may differ by food security status or have greater relevance than food security to dietary outcomes. Our food-secure and food-insecure groups were similar on health and sociodemographic variables, but may have differed on other factors, such as household income, which were not measured. On the community advisory board’s advice, we did not measure household income; however, all women were from high-poverty tracts. Low neighborhood SES is a significant predictor of negative health impacts, including weight gain among college-educated Black women.41 It is noteworthy that others have found adjustments for income and food assistance benefits in similarly disadvantaged samples produced more non-significant findings (or made no difference) for associations between food security and both diet quality and intake;18 thus, the fact that we could not adjust for household income or food assistance benefits is not a likely explanation for our findings of mainly non-significant differences between women from food-secure and food-insecure households. The relative importance of food security and income compared with other influences on dietary intake, such as taste preferences, socio-cultural foodways, and the neighborhood food environment, remains unsettled.

Conclusion

Our study contributes to the literature on food security and dietary intake by comparing women in food-secure and food-insecure households in a relatively homogeneous sample of overweight/obese, mainly African American women from neighborhoods of high poverty. There were few differences in dietary intake, contrary to our hypothesis. Compared with studies cited above, we found fewer statistically significant differences by food security status; however, differences of small magnitude (eg, .10 to .20 serving)22 between food security categories reach statistical significance in very large national samples. Our study confirms findings across other studies that poor nutrition is prevalent regardless of food security status, a profile that leads to elevated chronic disease risk and threatens effective chronic disease management, thereby contributing to health disparities.