The Association Between Anxiety Sensitivity and Food Cravings among Individuals Seeking Treatment for Weight-Related Behaviors

Abstract

Background:

A better understanding of risk-factors associated with state-like food cravings may be one clinically relevant component in an effort to better understand obesity. Existing work has shown anxiety sensitivity (AS) to be a significant risk factor for increased cravings across a variety of health behaviors (e.g., smoking, alcohol use). Yet, no work has examined the relationship between AS and state-like food cravings. Therefore, the current study sought to examine the association between AS and a variety of state-like food cravings, including: (1) an intense desire to eat, (2) anticipation of relief from negative states and feelings/improvement in mood that may result from eating, (3) obsessive preoccupation with food or lack of control over eating, and (4) craving as a physiological state.

Methods:

Participants included 161 (M_age = 31.58, SD = 10.71; 60.9% female) individuals seeking treatment for weight-related behaviors.

Results:

Results indicated that elevated AS was associated with reinforcement-based and physiological food cravings.

Conclusion:

Our findings indicate that there may be clinical utility in screening for AS among individuals seeking treatment for weight-related behaviors in efforts to better understand specific types of food craving.

1.0. Introduction

Obesity is a chronic and highly prevalent condition in the United States [U.S.; ¹]. It is estimated that 42.5% of adults in the U.S. meet criteria for obesity [2], and that number is expected to rise to 50% by 2030 [3]. Obesity is associated with an increased risk of developing heart disease, type 2 diabetes, stroke, and certain types of cancer, as well as increased risk for medical complications [e.g., increased risk for COVID-19 complications; ^{4, 5}]. To address the management and prevention of obesity, it is important to understand motivational processes that may be related to maladaptive behaviors that contribute to or maintain weight gain among individuals with obesity.

A better understanding of state-like food cravings may be one clinically relevant component in an effort to better understand obesity [6]. State-like food cravings is defined as an intense desire to consume a particular food item in response to momentary situations, or psychological and physiological states [7, 8]. Furthermore, food cravings often translate to consumption of craved food-items and is associated with overconsumption of food [binge eating; ⁹]. The most widely used measure of state-like food cravings is the Food Cravings Questionnaire-State (FCQ-S), which measures the intensity of momentary food cravings [7, 10]. The FCQ-S consists of five subscales, including: (1) an intense desire to eat; (2) anticipation of positive reinforcement that may result from eating; (3) anticipation of relief from negative states and feelings as a result of eating; (4) lack of control over eating; and (5) craving as a physiological state [i.e., hunger; ⁷]. Among individuals classified as overweight, state-like cravings are positively associated with self-reported sweet food and drink consumption and objective sweet food consumption [11]. Food cravings disrupt eating behaviors by making individuals more susceptible to external cues rather than internal hunger cues [12]. Thus, tactics aimed to reduce food cravings [e.g., cognitive strategies; ¹³] have been shown to assist in weight loss. Indeed, among individuals with obesity participating in a weight loss program, a reduction in state-like food cravings was positively associated with a greater percentage of weight loss [14]. Food cravings appear to be particularly relevant based on body mass index (BMI), with work suggesting higher BMI groups may be more susceptible to food intake in response to food cravings [15]. Notably, extant work has found state-like food cravings varies as a function of emotional eating [11], suggesting affective vulnerabilities may influence such variations in cravings.

An affective vulnerability factor that may be particularly relevant in influencing food cravings is anxiety sensitivity (AS). AS is a transdiagnostic cognitive vulnerability factor reflecting the fear of anxiety-related sensations (e.g., heart palpitations), which arises from beliefs that such symptoms will lead to harmful personal consequences [e.g., heart attack; ¹⁶]. AS demonstrates incremental validity relative to trait anxiety [17] and negative affectivity [18]. Further, AS is concurrently and prospectively associated with an increased risk of anxiety and mood psychopathology and more severe symptoms [19–21]. AS is a well-documented risk factor for increased cravings across a variety of health behaviors [e.g., smoking, alcohol use, marijuana; ^{22, 23, 24}]. Despite the promise of such work, no research has evaluated the relations between AS and state-like food cravings, specifically. Yet, studies have shown AS is related to several maladaptive eating processes and behaviors [e.g., eating expectancies, binge eating, greater calorie consumption; ^{25, 26–28}]. This data globally suggests that there is theoretical and clinical significance to better understanding AS-food cravings among affectively vulnerable individuals seeking treatment for weight-related behaviors.

Theoretically, AS may be associated with the presence of food cravings in an effort to counteract or regulate negative mood leading to an intense desire to eat or cravings motivated by affect regulation. In line with this perspective, extant work has found negative affect is related to an increased desire for food [e.g., chocolate; ^{29, 30}]. AS has been associated with eating cognitions and behaviors aimed at alleviating negative mood [25, 27, 31] suggesting AS may contribute to greater reinforcement-based food cravings. Additionally, loss of control of eating has been linked to difficulty coping with negative mood [32], suggesting AS may contribute to a lack of control over eating. AS may also be related to disrupted (stress-based) hormonal responses [15, 33]. Extant work has linked stress to increased ghrelin response (i.e., an appetite stimulus) which, in turn, was related to increased food cravings [15]. Interestingly, this association appears to vary based on BMI status, with higher BMI groups appearing more susceptible to stress-induced hormonal responses [15]. Thus, higher levels of AS may contribute to greater cravings at a physiological level (e.g., cravings to satisfy hunger).

The proposed study is a secondary analysis among individuals seeking treatment for weight-related behaviors interested in participating in a brief, single-session, computer-delivered AS reduction program [34]. The current study seeks to explore the association between AS and food cravings. The following food cravings processes were examined: (1) an intense desire to eat, (2) anticipation of relief from negative states and feelings/improvement in mood that may result from eating, (3) obsessive preoccupation with food or lack of control over eating, and (4) craving as a physiological state. It is hypothesized that anxiety sensitivity would be related to all food cravings subscales above and beyond the variance accounted for by age [35, 36] and negative affectivity [37, 38].

2.0. Method

2.1. Participants

The current study included 161 treatment-seeking participants (M_age = 31.58, SD = 10.71; 60.9% female) interested in participating in a larger study examining the impact of a computer-delivered intervention for people with obesity and elevated anxiety sensitivity [34]. Participants were invited to complete the baseline assessment based on pre-screen eligibility criteria, including having a BMI of 30 or higher calculated based on participant’s self-reported height and weight, self-report elevated AS defined as a score of 17 or greater on the Anxiety Sensitivity Index-3 [ASI; ^{39, 40}], and being at least 18 years of age. Participants were excluded at the pre-screen if they endorsed legal matters that they believed would interfere with their participation in all study components and for not being fluent in English. In the current study, participants reported an average BMI of 37.37 (SD = 6.01) and had an average AS score of 41.26 (SD = 14.12). The racial/ethnic distribution of the sample of the present study included 60.9% White (n = 98), 19.3% Black/African American (n = 31), 9.3% Asian (n = 15), 1.2% Native American/Alaska Native (n = 2), and 9.3% other (n = 15). Additionally, of the sample, 24.2% identified as Hispanic or Latino (n = 39).

2.2. Measures

Demographic Questionnaire.

A demographic questionnaire was used to collect data regarding gender, race/ethnicity, and age for descriptive purposes. Age will be utilized as a covariate in the current study.

Positive and Negative Affect Schedule [PANAS; ⁴¹].

The PANAS is a self-report measure that assesses the extent to which participants experienced 20 different feelings and emotions (e.g., excited, distressed) on a Likert-type scale ranging from 1 (Very slightly or not at all) to 5 (Extremely). The measure yields two factors, negative and positive affectivity. The PANAS has demonstrated strong psychometric properties, including high levels of internal consistency across clinical and non-clinical samples [41, 42]. The PANAS negative affectivity subscale (PANAS-NA) demonstrated excellent internal consistency (α = .91) in the current study and was used as a covariate

Anxiety Sensitivity Index-3 [ASI-3; ⁴⁰].

The ASI is an 18-item self-report measure of sensitivity to and fear of potential negative consequences (e.g., heart attack) of anxiety-related symptoms and sensations (e.g., heart palpitations). Participants are asked to rate on a 5-point Likert-type scale ranging from 0 (Very little) to 4 (Very much) the degree to which they are concerned about these possible negative consequences (range 0–72). The ASI-3 has sound psychometric properties, including excellent internal consistency, predictive validity, and reliability [40]. The ASI-3 demonstrated excellent internal consistency (α = .91) in the current study and was used as the predictor variable.

Food Cravings Questionnaire-State Version [FCQ-S; ⁷].

The FCQ-S is a 15-item self-report measure used to assess (1) an intense desire to eat (Food Craving-Desire); (2) anticipation of positive reinforcement that may result from eating (Food Craving- Positive Reinforcement); (3) anticipation of relief from negative states and feelings as a result of eating (Food Craving-Negative Reinforcement); (4) obsessive preoccupation with food or lack of control over eating (Food Craving-Control); and (5) craving as a physiological state (Food Craving-Physiological). Participants are asked to rate on a 5-point Likert-type scale the extent to which they agree with various food cravings statements ranging from 1 (Strongly disagree) to 5 (Strongly agree) for each item. As recommended by previous work, positive reinforcement- related items 4 and 6 were removed, and the subscales of Food Craving-Positive Reinforcement and Food Craving-Negative Reinforcement were combined to form one reinforcement scale [Food Craving-Reinforcement; ⁴³]. The FCQ-S has demonstrated strong psychometric properties in past work [10] and is validated among individuals classified as overweight or obese [43]. The Food Craving-Desire (α = .95), Food Craving-Reinforcement (α = .86), Food Craving-Control (α = .78), and Food Craving-Physiological (α = .83) subscales were used as criterion variables.

2.3. Procedure

Participants were recruited nationwide through flyers, electronic media (e.g., Facebook, Craigslist, listservs), and physician referrals. Participants in the current study completed a prescreen survey to determine initial eligibility. If found eligible at the prescreen, the participants were invited to complete a baseline assessment that further assessed their eligibility. If found ineligible at the baseline, participants were compensated $10 and given further referrals. If found eligible at the baseline, participants completed an online intervention and follow-up assessments. The study procedures for the parent study can be found elsewhere [34].

2.4. Analytic Strategy

Data analysis was conducted using SPSS 25.0. First, zero-order correlations and descriptive statistics on all study variables were examined. Then, the relationship between AS and four continuous criterion variables was examined via a 2-step hierarchical linear regression: (1) Food Craving-Desire, (2) Food Craving-Reinforcement, (3) Food Craving-Control, and (4) Food Craving-Physiological. Given past work has found food cravings decline with age [35, 36] and negative affectivity has contributed to greater maladaptive eating behaviors [37, 38], these variables were included as covariates in the current study. Step 1 of all models included covariates of age and negative affectivity. Step 2 included the addition of AS. Statistical significance was set at p ≤ .05. Using the F statistic, we evaluated model fit for each of the regressions. As a measure of effect size, we included squared semi-partial correlations [sr2; interpreted as .01 = small, .09 = moderate, and .25 = large; ⁴⁴]. The current study is a secondary analysis from a larger randomized-controlled trial in which the Power analysis to determine the appropriate sample size to detect effects was conducted [45].

3.0. Results

3.1. Descriptive Statistics and Bivariate Correlations

Table 1 presents zero-order correlations and descriptive statistics. AS evidenced a positive and statistically significant correlation with negative affectivity. Additionally, AS exhibited a positive and statistically significant association with the Food Craving-Reinforcement, Food Craving-Control, and Food Craving-Physiological subscales. Age yielded a negative and statistically significant association with all food cravings subscales. The Food Craving-Desire and Food Craving-Control subscales evidenced a positive and statistically significant correlation with negative affectivity. Additionally, all the food cravings subscales exhibited a positive and statistically significantly correlation with each other.

Table 1.

Descriptive statistics and bivariate correlations between study variables (N =161)

	1	2	3	4	5	6	7
1. Age	-
2. Negative Affectivity	−.08	-
3. Anxiety Sensitivity	−.12	.43***	-
4. Food Craving-Desire	−.30***	.17*	.14	-
5. Food Craving-Reinforcement	−.26***	−.01	.23**	.62***	-
6. Food Craving-Control	−.23**	.26***	.25**	.66***	.59***	-
7. Food Craving-Physiological	−.25**	−.01	.23**	.54***	.75***	.58***	-
Mean	31.58	27.33	41.26	11.34	13.69	10.55	10.04
SD	10.71	9.42	14.12	3.60	4.35	3.22	3.62
Possible Range	18–75	10–50	0–72	3–15	4–20	3–15	3–15

Note.

^*p < .05,

^**p < .01

^***p < .001;

Negative Affectivity = Positive and Negative Affect Schedule Negative Affect Score [41]; Anxiety Sensitivity = Anxiety Sensitivity Total Score [40]; Food Craving-Desire = Food Cravings Questionnaire-State Version An Intense Desire to Eat Subscale [7]; Food Craving-Reinforcement = Food Cravings Questionnaire-State Version Anticipation of Positive & Negative Reinforcement That May Result From Eating Subscale [7]; Food Craving-Control = Food Cravings Questionnaire-State Version Lack of Control Over Eating Subscale [7]; Food Craving-Physiological = Food Cravings Questionnaire-State Version Craving as Physiological State Subscale [7].

3.2. Regression Analyses

Table 2 presents hierarchal regression results. For the Food Craving-Desire subscale, step 1 with covariates was statistically significant (R² = .11, F(2, 158) = 9.76, p < .001); age was a statistically significant predictor. In step 2, AS was added and the model remained statistically significant (R² = .11, F(3, 157) = 6.65, p < .001). However, the addition of AS in step 2 did not account for a statistically significant increase in variance in the Food Craving-Desire subscale (ΔR² = <.01, F(1, 157) = .49, p = .484).

Table 2.

Hierarchical Regression Results

Food Craving-Desire
Model		b	SE	β	t	p	CI (l)	CI (u)	sr 2
1	Age	−0.10	0.03	−0.29	−3.81	< .001	−0.15	−0.05	.082
	Negative Affectivity	0.06	0.03	0.15	1.93	.055	0.00	0.11	.021
2	Age	−0.09	0.03	−0.28	−3.72	< .001	−0.15	−0.04	.078
	Negative Affectivity	0.05	0.03	0.12	1.45	.150	−0.02	0.11	.012
	Anxiety Sensitivity	0.01	0.02	0.06	0.70	.484	−0.03	0.06	.003
Food Craving-Reinforcement
Model		b	SE	β	t	p	CI (l)	CI (u)	sr 2
1	Age	−0.11	0.03	−0.26	−3.39	< .001	−0.17	−0.04	.068
	Negative Affectivity	−0.02	0.04	−0.03	−0.44	.658	−0.09	0.05	.001
2	Age	−0.10	0.03	−0.24	−3.17	.002	−0.16	−0.04	.056
	Negative Affectivity	−0.07	0.04	−0.15	−1.76	.081	−0.14	0.01	.017
	Anxiety Sensitivity	0.08	0.03	0.26	3.16	.002	0.03	0.13	.056
Food Craving-Control
Model		b	SE	β	t	p	CI (l)	CI (u)	sr 2
1	Age	−0.06	0.02	−0.21	−2.83	.005	−0.11	−0.02	.045
	Negative Affectivity	0.08	0.03	0.24	3.22	.002	0.03	0.13	.058
2	Age	−0.06	0.02	−0.20	−2.67	.008	−0.10	−0.02	.040
	Negative Affectivity	0.06	0.03	0.18	2.20	.029	0.01	0.12	.027
	Anxiety Sensitivity	0.03	0.02	0.14	1.73	.086	0.00	0.07	.017
Food Craving-Physiological
Model		b	SE	β	t	p	CI (l)	CI (u)	sr 2
1	Age	−0.09	0.03	−0.26	−3.3	.001	−0.14	−0.03	.065
	Negative Affectivity	−0.01	0.03	−0.03	−0.42	.678	−0.07	0.05	.001
2	Age	−0.08	0.03	−0.23	−3.08	.002	−0.13	−0.03	.053
	Negative Affectivity	−0.06	0.03	−0.14	−1.73	.085	−0.12	0.01	.017
	Anxiety Sensitivity	0.07	0.02	0.26	3.17	.002	0.03	0.11	.056

Note. N = 161.

^*p < .05,

^**p < .01

^***p < .001;

Negative Affectivity = Positive and Negative Affect Schedule Negative Affect Score [41]; Anxiety Sensitivity = Anxiety Sensitivity Total Score [40]; Food Craving-Desire = Food Cravings Questionnaire-State Version An Intense Desire to Eat Subscale [7]; Food Craving-Reinforcement = Food Cravings Questionnaire-State Version Anticipation of Positive & Negative Reinforcement That May Result From Eating Subscale [7]; Food Craving-Control = Food Cravings Questionnaire-State Version Lack of Control Over Eating Subscale [7]; Food Craving-Physiological = Food Cravings Questionnaire-State Version Craving as Physiological State Subscale [7].

In regards to the Food Craving-Reinforcement subscale, step 1 with covariates was statistically significant (R² = .07, F(2, 158) = 5.76, p = .004). Age was a statistically significant predictor of the Food Craving-Reinforcement subscale. AS was added in step 2 and the model remained statistically significant (R² = .12, F(3, 157) = 7.39, p < .001) and accounted for a statistically significant increase in variance in the Food Craving-Reinforcement subscale (ΔR² = .06, F(1, 157) = 10.00, p = .002). Age and AS were statistically significant predictors.

In the Food Craving-Control subscale regression analysis, step 1 with only covariates was statistically significant (R² = .11, F(2, 158) = 9.96, p < .001); age was a statistically significant predictor. When AS was added in step 2, the model remained statistically significant (R² = .13, F(3, 157) = 7.72, p < .001), although the addition of AS did not account for a statistically significant increase in variance in the Food Craving-Control subscale (ΔR² = .02, F(1, 157) = 2.99, p = .086).

Lastly, for the Food Craving-Physiological subscale, step 1 with covariates was statistically significant (R² = .07, F(2, 158) = 5.47, p = .005); age was a statistically significant predictor. In step 2, AS was added and the model remained statistically significant (R² = .12, F(3, 157) = 7.20, p < .001), accounting for a statistically significant increase in variance in the Food Craving-Physiological subscale (ΔR² = .06, F(1, 157) = 10.03, p = .002). Age and AS were statistically significant predictors of the Food Craving-Physiological subscale.

4.0. Discussion

The results of the current study indicated that greater levels of AS were associated with greater levels of: (1) anticipation of relief from negative states and feelings/improvement in mood that may result from eating and (2) craving as a physiological state. These results were evident after accounting for theoretically relevant covariates of age and negative affectivity. The size of the incremental effects of AS were small (6% of variance for each criterion variable). The current study is in line with previous work suggesting the association of AS with cravings across other health-related behaviors [e.g., smoking, alcohol use, marijuana; ^{22, 23, 24}] and eating-related outcomes [e.g., eating expectancies, binge eating, greater calorie consumption; ^{25, 26–28}] and extends this work to highlight the role of AS in state-like food cravings.

Drawing from the current results, it may be theorized that individuals with higher levels of AS may experience increased food cravings guided by positive or negative reinforcement motives in an effort to downregulate negative mood [25, 27, 31]. Moreover, AS may also trigger a ghrelin response [15] and individuals with elevated AS may be more hypervigilant to internal experiences (e.g., hunger) triggering greater cravings as a physiological state (e.g., hunger). These maladaptive motivational processes are likely to be associated with more imminent alleviation of negative symptoms in spite of the long-term consequences [22]. Yet, over time, efforts to gain short-term relief (e.g., through eating) may result in additional negative experiences [e.g., chronic pain associated with weight gain, increased hunger; ^{46, 47}] which may further maintain and/or exacerbate maladaptive motivational processes such as food cravings.

Contrary to our hypothesis, AS was not a significant predictor of: (1) an intense desire to eat or (2) obsessive preoccupation with food or lack of control over eating. These findings may be due to the fact that AS may be more theoretically relevant to cravings motivated by an internal experience (e.g., feelings of weakness, irritability) rather than more general desires that are not linked to an internal experience (e.g., intense desire to eat more generally). Moreover, AS does not appear to be particularly relevant as it relates to cravings guided by obsessive preoccupations with or inability to control the intake of certain types of food. Similarly, this lack of association may be due to the context guiding the cravings. For example, unlike other subscales (e.g., reinforcement and physiological), obsessive preoccupation with food or lack of control over eating cravings were not specifically tied to cravings motivated by an internal experience (e.g., improving mood, reducing sensations associated with hunger).

Although not a primary aim, it is important to highlight that our findings indicated older age was a significant predictor of reduced levels of food cravings across all four craving subscales. This finding is in line with existing work suggesting levels of food cravings decline with age [35, 36]. Theoretically, as individuals age, they may experience a decline in their dopamine receptors [48]. Therefore, certain foods may no longer trigger the same dopamine response, reducing food cravings over time [49]. Moreover, individuals may experience a decline in appetite as they age [50], reducing the intensity and desire for certain foods. Future work is needed to identify for whom interventions targeting food cravings are most appropriate, thereby enhancing the potential for precision medicine among individuals seeking treatment for weight-related behaviors.

Results of the current study highlight motivational processes (i.e., increased AS) that may be guiding reinforcement and physiological-oriented food cravings. As such, screening for AS among individuals seeking treatment for weight-related behaviors may be useful in guiding case conceptualization. For example, interoceptive exposure may have clinical utility in reducing AS [51], which, in turn, may contribute to decreased reinforcement-based food cravings (e.g., eating would improve my mood). Emerging work has examined interoceptive exposure for eating-related behaviors with encouraging results [52]. Moreover, individuals with elevated AS seeking treatment for weight management may have increased hypervigilance to internal hunger cues (e.g., dizziness). As such, interventions aimed to reduce the fears associated with such internal experiences (e.g., I worry if I feel dizzy, I might pass out) may be useful in reducing cravings motivated by physiological states (e.g., hunger).

The study has several limitations. First, measures were collected via self-report. Thus, the findings could be influenced by shared method variance. Second, the current study had a cross-sectional design utilizing data from a larger longitudinal study. Future longitudinal studies are needed to establish temporal precedence and directionality of the observed relations. Third, although we had a fairly diverse sample, the majority of the sample was White (60.9%). Thus, it is important for future studies to examine the proposed relations amongst a more diverse and representative sample to determine if there are differences based on race/ethnicity as well as other sociodemographic variables (e.g., gender). Finally, the current study examined participants seeking treatment for weight-related behaviors. Further research may be needed to examine the impact of food cravings and affective vulnerabilities on non-treatment-seeking individuals at risk for weight-related problems.

Overall, the current study provides initial evidence of the role of AS and reinforcement-based (e.g., positive and negative) cravings as well as cravings as a physiological state. Future longitudinal research is needed to further explicate and validate the observed associations to better understand the clinical importance of AS in terms of specific types of food cravings. Additionally, future work could further expand on the proposed model and examine the impact of AS on food cravings and subsequent weight-related outcomes.

Highlights:

• State-like food cravings may be relevant in the maintenance of obesity.

• AS is associated with reinforcement-based and physiological food cravings.

• Older age is associated with reduced food cravings.

• Screening for AS in weight management interventions may have clinical utility.