Food cravings are complex psychological phenomena characterized as frequent, intense desires to consume a specific type of food that is often difficult to resist. The Food Craving Inventory (FCI) is a validated self-report questionnaire designed to quantify the frequency of food cravings along five dimensions: overall food cravings (FCI-O); cravings for sweets (FCI-S); cravings for high-fat food (FCI-H); cravings for starchy food (FCI-St); and cravings for fast food (FCI-FF) [1]. Specific food craving categories, as measured by the FCI, are positively associated with food intake of the same category at a meal [2, 3], as well as habitual intake of food of that type [4]. The construct validity of the FCI is well established in studies demonstrating positive associations between scale-specific cravings and specific food intake [2, 5, 6].

Several potential neurobiological mechanisms underlying these hunger-craving associations are suggested by the literature. Extended fasting periods have been shown to modulate neural responses to food cues in reward-related brain regions [7, 8]. Moreover, an objective measure of homeostatic hunger (i.e., plasma insulin) was more closely aligned with neural activation of reward-related brain regions than subjective hunger ratings [9]. Furthermore, changes in ghrelin have been positively correlated with a six-item FCI “state” scale (pizza, brownie, chocolate, cookies, ice cream, steak), and higher baseline ghrelin is predictive of longitudinal increases in FCI-St and FCI-O after adjusting for demographic variables, baseline BMI, and baseline food cravings [10]. Finally, recent studies have shown that treatments involving tirzepatide, a GIP/GLP-1 analog, reduce FCI-O, FCI-S, and FCI-St in people with obesity [11]. These findings collectively suggest potential mechanistic underpinnings of FCI outcomes.

It is in this that we note two significant limitations in the current FCI literature. First, the majority of clinical studies that use the FCI, in their attempts to control for the influences of hunger, rely on self-reported hunger (i.e., visual analog scales; VAS) [12, 13] rather than objective, behaviorally defined measures, which may be more appropriate in the context of the neurohormonal influences described above. This is further supported by the fact that self-reported hunger reflects the psychological drive to eat, which is distinct from food cravings and physiological hunger [13]. Together, these findings suggest that the typical practice of using VAS to measure hunger [14] may not be optimal in light of the likely impact of objective hunger states on the FCI. This further suggests that the use of objectively measured, behaviorally defined hunger (hours since last caloric intake; FAST) may be preferable. Further support for this notion can be found elsewhere in the literature. Objective hunger (i.e., a 4-h fast) has been found to affect image-induced food cravings [15], and previous studies have suggested that food cravings may develop as a result of learned associations between ingesting specific foods and hunger states (a 2-h fast vs. a fed state) [16]. Importantly, objective hunger (i.e., an 18-h fast) increases neural activity response to high-calorie foods [17], suggesting that hunger may increase the likelihood of energy-dense food cravings. Additional evidence further suggests that objective hunger (i.e., a 17-h fast) can make food cravings more difficult to resist [18]. Despite these findings, most FCI studies either neglect to quantify current hunger or rely on unreliable [19] self-report modalities (e.g., VAS).

In summary, the convergence of evidence suggests that hunger may modulate FCI scores and that to accurately capture hunger as a covariate in FCI studies, behaviorally defined, objectively measured hunger (FAST) may be the most appropriate method.

Another limitation in the current literature is that few studies adequately account for menstrual cycle phase, despite its known influences on subjective food cravings. Neuroimaging studies have found that the phase of the menstrual cycle can have salient influences on food cue reactivity in premenopausal women as measured by functional magnetic resonance imaging (fMRI-FCR) [20, 21], where incidental characteristics are crucial to control for [22]. Furthermore, certain phases of the menstrual cycle have been associated with consumption of sweet, carbohydrate, and fatty foods [23, 24]. Together, these suggest that it is imperative to control for menstrual cycle influences when measuring food cravings with the FCI.

A final and important consideration when using the FCI lies in the interpretation of the FCI as representing an enduring pattern in the relationship with food, versus a more transient, situationally specific affinity [21]. Evidence supporting the FCI as a reflection of an enduring pattern that contributes to longer-term weight gain via chronic over-ingestion patterns remains equivocal. In one cross-sectional study, each of the FCI subscales was found to have weak, positive associations with BMI (FCI-S: r = 0.13, p < 0.05; FCI-H: r = 0.21, p < 0.001; FCI-St: r = 0.15, p < 0.001; FCI-FF: r = 0.13, p < 0.05) [4]. However, this study had a large sample size (n = 646); and the variance of BMI accounted for by each of the FCI constructs was low (FCI-S: 1.7%; FCI-H: 4.4%; FCI-St: 2.3%; FCI-FF: 1.7%), suggesting a relatively low effect size compared to other, more salient predictors. In another cross-sectional worksite study, it was found that FCI-O was associated with BMI (r = 0.21, p = 0.046), with similarly weak associations noted [25]. The variance explained was similarly less than 5%. Together, these findings do not present compelling evidence that food cravings, as measured by the FCI, represent an enduring propensity towards ingestion based on the presence of food cravings.

In summary, while the FCI is widely used in the obesity and ingestive behavior literature to better understand the psychological and neurophysiological response to dietary interventions [26, 27], there are methodological considerations that warrant attention. Here we seek to examine (1) the potential association of objective, behaviorally measured hunger (FAST) with responses on the FCI, controlling for the influence of the menstrual cycle and (2) the relationships of the FCI with weight-related measures (i.e., BMI, BW, BF) in a treatment-seeking population with class I & II obesity.