Abstract
Objective
Food noise has received attention in the media, although no validated questionnaires exist to measure it. This study developed and tested the reliability and validity of the Food Noise Questionnaire (FNQ).
Methods
Participants (N = 400) successfully completed, the FNQ and a demographic questionnaire and self‐reported weight and height. A subsample (n = 150) completed the FNQ 7 days later for test–retest reliability, and this subsample's first FNQ data were subjected to exploratory factor analysis. The remaining subsample (n = 250) completed two preoccupation with food questionnaires to test convergent validity, along with mood, anxiety, and stress questionnaires to test for discriminant validity. Confirmatory factor analysis was conducted using this subsample's FNQ data.
Results
Data from 396 participants were analyzed (4 participants did not complete all FNQ items). The FNQ had excellent internal consistency reliability (Cronbach α = 0.93) and high test–retest reliability (r = 0.79; p < 0.001; mean [SD] = 7.4 [1.0] days between administration). Factor analyses found that the five FNQ items loaded onto a single factor, with good fit indices (χ 2[5] = 52.87, p < 0.001; root mean square error of approximation [RMSEA] = 0.20; comparative fit index [CFI] = 0.95; standardized root mean squared residual [SRMR] = 0.03). The FNQ showed good convergent (all r > 0.78; p < 0.001) and discriminant (all r < 0.39; p < 0.001) validity.
Conclusions
The FNQ provides a psychometrically reliable and valid measure of food noise, although further research is needed to evaluate its clinical utility.
Study Importance.
What is already known?
The construct of food noise has garnered significant attention, with anecdotal reports from patients demonstrating that recently approved weight loss drugs greatly diminish food noise.
No validated questionnaires exist to measure or quantify food noise.
What does this study add?
The current study, to our knowledge, is the first to develop and validate a tool to measure food noise.
The results indicate that the Food Noise Questionnaire (FNQ) is a reliable and valid tool to assess food noise.
How might these results change the direction of research or the focus of clinical practice?
The FNQ provides a brief (five‐item) practical tool for researchers and clinicians to measure food noise.
The FNQ provides a psychometrically valid tool to quantify food noise and its correlates in various populations. Future use of the tool among different populations and during weight loss treatment is needed to determine the clinical utility of the construct and questionnaire.
INTRODUCTION
The concept of “food noise” has received increased attention since the introduction of glucagon‐like peptide‐1 receptor agonists (GLP‐1) to treat obesity in the early 2020s [1]. Based on anecdotal reports from clinical settings, patients describe food noise as the experience of having a constant inner dialogue about food [2] and a continuous flux of food‐related thoughts that include thinking about what to eat, how much to eat, and when to eat again [3]. Patients who are treated with GLP‐1 receptor agonists and similar compounds often report a noticeable reduction in food‐related thoughts [4]. Despite a high level of interest in food noise, there are no psychometrically valid measures of the construct. Therefore, in the current study, we developed and tested the reliability and validity of a questionnaire to measure food noise.
To date, one publication has reviewed food noise from a scientific viewpoint, defining the concept as “heightened and/or persistent manifestations of food cue reactivity, often resulting in intrusive food thoughts and maladaptive eating behaviors” [5]. This definition provides an excellent overview of the concept; however, it views food noise primarily as a food cue reactivity phenomenon, which does not entirely reflect peoples' reports of food noise. Specifically, patients report experiencing food noise even without food cues [6], although it is recognized that food noise may indeed be triggered or exacerbated by food cues. Patients also report that a distinctive feature of food noise is that the thoughts are persistent and often intrusive [2]. Finally, patients report that food noise negatively affects their quality of life and makes it difficult to engage in healthy lifestyle behaviors [6]. Accordingly, we defined food noise as persistent, intrusive thoughts about food that are disruptive to daily life and make healthy behaviors difficult.
A reliable and valid measure of food noise is needed that can be used in clinical and research settings. The purpose of the current study was to develop and assess the reliability and validity of a food noise questionnaire. We hypothesized that the questionnaire would have adequate test–retest and internal consistency reliability and that the items would load onto a single factor. We further hypothesized that the convergent and discriminant validity of the questionnaire would be supported.
METHODS
Food noise definition, item generation, and response format
In order to define food noise, we reviewed the literature, patient reports in various social media outlets, and direct conversations with patients with lived experience. Items for the Food Noise Questionnaire (FNQ) were then created to reflect this definition. The response format for the FNQ is a 5‐point Likert‐type scale: strongly disagree (scored as 0); disagree (scored as 1); neither agree nor disagree (scored as 2); agree (scored as 3); and strongly agree (scored as 4). A single total score for the questionnaire is calculated by summing responses for the five items, and higher total scores indicate greater levels of food noise. The item list of the FNQ is provided in Table S1. The questionnaire instructions were as follows: “Please answer these questions by reflecting on your thoughts over the last 2 weeks only.” A 2‐week recall period was used to ensure that participants' ratings were not influenced by brief situational experiences that were associated with high or low levels of food noise.
Eight candidate items for the FNQ were created and distributed to nine experts, who evaluated their content validity. The nine experts had expertise in survey development and validation, weight management, obesity medicine, eating behavior, nutrition, and metabolism. In order to maintain the independence of experts, they conducted their reviews individually, and the identities of the other experts were not revealed to any expert aside from the authors [7]. In line with recommendations [7], the experts were provided with a standardized form to provide their opinion on the instructions, individual items, and response options of the initial questionnaire [8]. The experts were also asked to provide written comments with their suggestions and proposed edits to items. The standardized form is provided in Table S2. Upon receiving experts' feedback, our research team evaluated each item for clarity and redundancy and incorporated the experts' suggested edits.
Participants
Participants were recruited from a third‐party vendor's participant pool of individuals residing in various locations in the United States (EMI Research Solutions). A total sample size of 400 participants successfully completed the online survey. A subsample (n = 150) completed the FNQ 7 days later to evaluate test–retest reliability. The remaining subsample (n = 250) answered additional questionnaires to test convergent and discriminant validity. The sample sizes were developed considering that the instrument would consist of no more than 10 items. Therefore, we used a sample size consistent for questionnaire development, namely that the respondent to item ratio be at least 10:1 [9].
Participants aged 18 years or older were eligible for participation in the study. The study excluded participants who were participating in any weight loss programs (e.g., participating in a structured weight loss or lifestyle change program), were using prescription or over‐the‐counter antiobesity medications, or self‐reported having a body mass index (BMI) lower than 18.5 kg/m2. Our goal was to recruit a nationally representative sample of participants in line with the latest Centers for Disease Control and Prevention (CDC) [10] and US Census Bureau report [11] for BMI, sex, and race and ethnicity. Specifically, we aimed to recruit approximately one‐half male participants, 31% with overweight (BMI between 25 and 29.9 kg/m2), 42% with obesity (BMI > 30 kg/m2), and 9% with severe obesity (BMI > 40 kg/m2) and a sample that is approximately 60% White, 18% Hispanic or Latino, 12% Black, 5% Asian, and 5% other races.
Reliability tests
Internal consistency reliability (Cronbach α) was calculated to evaluate the consistency of the responses across FNQ items. The results of the descriptive statistics and inter‐item correlations were also used for item evaluation. Test–retest reliability was assessed via Pearson correlation [12]. A 7‐day time interval was used, given that a shorter time interval can cause carry‐over effects due to memory, whereas a longer interval can increase the likelihood of a change in the status of the construct [13].
Validity tests
An exploratory factor analysis was performed to identify the potential underlying factor structure of the FNQ. Subsequently, a confirmatory factor analysis was conducted to verify the factor structure. In order to assess convergent validity, we used two validated questionnaires that assessed individuals' preoccupation with food, which is hypothesized to be related to the food noise construct. Specifically, we compared the FNQ with the score from the three‐item Frequency subscale of the Food Preoccupation Questionnaire [14]. The Food Preoccupation Questionnaire has high internal consistency and test–retest reliability and has been validated in various samples. The Frequency subscale evaluates the extent to which participants thought about food. Thus, strong correlations between this subscale of the Food Preoccupation Questionnaire and the FNQ would support the convergent validity of the FNQ. In order to further examine convergent validity, we compared participants' FNQ scores with the score from the Preoccupation with Food subscale of the Food Cravings Questionnaire‐Trait (FCQ‐T) [15]. The FCQ‐T has been tested in multiple samples and has proven to be reliable and valid [16]. The Preoccupation with Food subscale consists of seven items measuring levels of food preoccupation.
Based on the literature, a robust test of discriminant validity involves the use of similar but theoretically distinct constructs [17]. Therefore, we used three validated questionnaires that assessed individuals' mood, anxiety, and stress, representing constructs that may be related but theoretically distinct from the food noise construct. We compared the FNQ with the score from the Patient Health Questionnaire (PHQ‐8) [18]. The PHQ‐8 comprises the first eight items of the nine‐item PHQ (PHQ‐9), and it does not include the item that assesses suicidal ideation and intent [19]. The PHQ‐8 is widely used in many countries for screening and assessing depression severity [20]. Additionally, the FNQ was compared with the 10‐item Perceived Stress Scale (PSS) [21]. The PSS measures the degree to which situations in one's life are considered stressful. Finally, the FNQ was compared with the General Anxiety Disorder‐7 (GAD‐7) instrument [22]. The GAD‐7 is a seven‐item self‐report questionnaire that assesses anxiety symptoms over the past 2 weeks. In order to support the discriminant validity, we expected lower correlations between the FNQ and the PHQ‐8, PSS, and GAD‐7 compared with the correlations between the FNQ and the Frequency subscale of the Food Preoccupation Questionnaire and the Preoccupation with Food subscale from the FCQ‐T.
Procedures
Participants were invited via email to complete an online survey created using Qualtrics software. On the first page of the survey, participants electronically consented to the study by selecting a response after reviewing an approved consent form. In order to confirm their eligibility for the study, participants answered initial eligibility questions, and those who were not eligible were screened out before answering any further questions. All eligible participants reported their education level, household income, employment status, age, sex, race and ethnicity, height, weight, weight history (e.g., history of weight gain or loss, diet attempts, weight fluctuations, disordered eating), medical conditions, and medication usage. After participants completed the initial demographic information, the whole sample (N = 400) completed the FNQ. Then, participants were randomly assigned to complete the FNQ 7 days later to evaluate test–retest reliability (n = 150) or to complete additional questionnaires to assess convergent and discriminant validity (n = 250). The randomization process was performed using the randomizer option in Qualtrics software, which used a nonstratified random sampling method. Respondents were required to complete a minimum of 95% of the survey items on each survey screen to successfully complete the online survey and receive payment. Participants received $25 upon completion of the survey. The survey was administered using a proprietary algorithm involving ~32 digital fingerprints. This algorithm was designed to ensure that participants did not retake the survey and to prevent bots from completing the survey. All study procedures were approved via expedited review by the Institutional Review Board at Pennington Biomedical Research Center. The study is registered at ClinicalTrials.gov (NCT06315907).
Statistical analysis
Internal consistency reliability was assessed using Cronbach α with the entire dataset. Test–retest reliability was assessed using Pearson correlation among the subsample that was asked to complete the instrument ~1 week after the first assessment. Correlations greater than 0.70 were considered satisfactory for both internal consistency and test–retest reliability [23]. An exploratory factor analysis was conducted on the test–retest sample, and a confirmatory factor analysis was conducted on the sample for convergent and discriminant validity. Model fit was evaluated by the model χ2 [24], the root mean square error of approximation (RMSEA) [25], the comparative fit index (CFI) [26], and the standardized root mean squared residual (SRMR) [27]. Chi‐square (χ2) values with p > 0.05 and CFI > 0.90 indicate a better fit [24, 26]. For the RMSEA, values < 0.08 indicate an adequate fit, and for the SRMR, values < 0.05 indicate a very good fit [25, 27].
Convergent and discriminant validity was assessed using Pearson correlation among the 250 participants who completed the additional questionnaires. For convergent validity, we compared the FNQ to the three‐item Frequency subscale of the Food Preoccupation Questionnaire and the Preoccupation with Food subscale of the FCQ‐T. Discriminant validity was tested by evaluating the relationships of the total score of the FNQ and the total scores of the PHQ‐8, the PSS, and the GAD‐7. Fisher's z‐transformation was used to calculate 95% confidence intervals (CI) to compare the correlations between the FNQ and the questionnaires used for convergent versus discriminant validity. Differences in FNQ scores by sex, race and ethnicity, and weight status were examined using ANOVA. Linear regression was also used to examine the relationship between the total score of the FNQ and participants' BMI value. In order to check the quality of the survey data, we examined the distribution of the data for potential outliers or erratic response patterns. Continuous outcomes are reported as mean ± SD. Data were analyzed using SAS software version 9.4 (SAS Institute Inc.) and Stata Statistical Software release 18 (StataCorp LLC).
RESULTS
Content and face validity
As a result of the expert review, five out of nine experts considered one item (i.e., “My thoughts about food interfere with what I need to do”) to be inadequate for assessing the food noise construct; therefore, the item was not included in the original seven‐item questionnaire (Table 1). Additionally, experts noted that both the instructions and response options included in the FNQ were applicable to the target population. The language on six out of the seven items was revised based on expert reviews to better reflect the food noise construct and improve clarity. The revisions to the language of the FNQ improved the readability, and the Flesch–Kincaid grade level [28] for the final version was 4.6 compared with 5.6 for the initial version.
TABLE 1.
Initial seven items and response options in the FNQ.
| Items |
|---|
| 1. I find myself constantly thinking about food throughout the day. a |
| 2. I seem to think about food all day, even though I do not want to. b |
| 3. My thoughts about food feel uncontrollable. |
| 4. I'm always thinking about food, even when I'm not hungry. b |
| 5. I spend too much time thinking about food. |
| 6. My thoughts about food have negative effects on me and/or my life. |
| 7. My thoughts about food distract me from what I need to do. |
Abbreviation: FNQ, Food Noise Questionnaire.
The response format for the FNQ is a 5‐point Likert‐type scale: strongly disagree (scored as 0); disagree (scored as 1); neither agree nor disagree (scored as 2); agree (scored as 3); and strongly agree (scored as 4).
These two items were excluded from the final questionnaire because they were highly correlated with at least one other item and were redundant. The second item was correlated with the first item, whereas the fourth item was correlated with the fifth item (data not shown).
Sample characteristics
The study screened 772 individuals, of whom 412 were eligible and enrolled/started the online survey. After the initial demographic questionnaire, a total of 409 participants were randomized to test–retest reliability and construct validity groups (3 participants failed to complete the required number of items in demographic questionnaire). A total of 400 participants successfully completed the online survey, and the 396 who completed all FNQ items were included in the current analysis. A detailed participant flow diagram is included in Figure S1. Based on the quality check of the survey data, no outliers or erratic values were identified; therefore, no data points were excluded from further analysis. Most participants were non‐Hispanic White (286, 72%) and female (266, 67%). Mean (SD) age and BMI values were 51.5 (13.6) years and 31.0 (7.1) kg/m2, respectively. Under one‐third (28.5%) of participants reported that they were currently dieting for weight loss. The characteristics of the survey sample are shown in Table 2.
TABLE 2.
Characteristics of the study sample (n = 396)
| n (%) | |
|---|---|
| Sex | |
| Male | 130 (32.8) |
| Female | 266 (67.2) |
| Race | |
| White/non‐Hispanic | 286 (72.2) |
| Black or African American/non‐Hispanic | 48 (12.1) |
| Hispanic | 25 (6.3) |
| Asian | 11 (2.8) |
| Other or prefer not to answer | 26 (6.6) |
| Ethnicity | |
| Hispanic | 25 (6.3) |
| Non‐Hispanic | 361 (91.2) |
| Prefer not to answer | 10 (2.5) |
| BMI, kg/m2 | |
| 18.5–24.9 | 72 (18.2) |
| 25–29.9 | 122 (30.8) |
| 30–39.9 | 166 (41.9) |
| >40 | 36 (9.1) |
| Dieting to lose weight | |
| Yes | 113 (28.5) |
| No | 283 (71.5) |
| Age category, y | |
| 18–34 | 45 (11.4) |
| 35–54 | 195 (49.2) |
| >55 | 156 (39.4) |
| Education a | |
| Some high school | 11 (2.8) |
| High school diploma or GED | 81 (20.5) |
| Some college | 148 (37.4) |
| Bachelor's degree | 106 (26.8) |
| Postgraduate degree | 50 (12.6) |
| Income a | |
| Less than $50,000 | 159 (40.2) |
| $50,000–$69,999 | 69 (17.4) |
| $70,000 and above | 161 (40.7) |
| Prefer not to answer | 7 (1.8) |
| Employment a | |
| Unemployed | 49 (12.4) |
| Full‐time employment | 161 (40.7) |
| Part‐time employment | 53 (13.4) |
| Retired | 89 (22.5) |
| Other | 44 (11.1) |
| Mean ± SD (range) | |
|---|---|
| Age, y | 51.5 ± 13.6 (20.0–88.0) |
| Height, cm | 168.7 ± 10.2 (134.6–205.7) |
| Weight, kg | 88.1 ± 21.4 (43.5–172.4) |
| BMI, kg/m2 | 31.0 ± 7.1 (18.5–60.4) |
Abbreviation: GED, General Educational Development.
All percentages may not total 100% due to rounding.
Reliability
The initial seven‐item FNQ had a high overall internal consistency reliability (Cronbach α = 0.95). The item‐to‐item correlations indicated that two items were highly correlated with at least one other item and were redundant [29]; therefore, the two items were excluded from the final questionnaire (i.e., “I seem to think about food all day, even though I do not want to” and “I'm always thinking about food, even when I'm not hungry”). The final five‐item FNQ had high internal consistency reliability (Cronbach α = 0.93). Pearson correlation among the five items ranged from r = 0.61 to r = 0.81. The total FNQ score had a mean (SD) of 7.39 (5.37) with a median of 6.0. The total score ranged from 0 to 20, with a higher score indicating greater food noise. Among the participants who were randomly assigned to complete the FNQ 7 days later, 148 answered all FNQ questions at baseline, and 137 completed the FNQ at follow‐up (mean [SD] = 7.4 [1.0] days). Test–retest reliability from these 137 participants was high (r = 0.79; p < 0.001).
Validity
The exploratory factor analysis (n = 148) with the final five‐item version of the FNQ indicated that all items loaded onto a single factor, with factor loadings ranging from 0.72 to 0.89 (Table 3). The item‐total correlations for each of the five items on the total score ranged from 0.79 to 0.91. The confirmatory factor analysis (n = 248) with the final five‐item version of the FNQ supported the one‐factor structure (χ 2[5] = 52.87, p < 0.001; RMSEA = 0.20; CFI = 0.95; SRMR = 0.03). The indices for CFI and SRMR suggested a good overall fit. The χ 2 test was significant, but this test is sensitive to large sample size, and RMSEA has been shown to have limitations for models with few degrees of freedom, as occurs with a single‐factor instrument [30]. The results were similar for the seven‐item version for both exploratory (overall factor loading of 0.95) and confirmatory factor analysis (χ 2[14] = 148.16, p < 0.001; RMSEA = 0.20; CFI = 0.93; SRMR = 0.04).
TABLE 3.
Factor loadings and item statistics of the final FNQ from the exploratory factor analysis (n = 148)
| Items | Original item no. | Mean (SD) | Factor loading |
|---|---|---|---|
| I find myself constantly thinking about food throughout the day | 1 | 1.91 (1.17) | 0.72 |
| My thoughts about food feel uncontrollable | 3 | 1.45 (1.19) | 0.87 |
| I spend too much time thinking about food | 5 | 1.70 (1.29) | 0.89 |
| My thoughts about food have negative effects on me and/or my life | 6 | 1.57 (1.27) | 0.87 |
| My thoughts about food distract me from what I need to do | 7 | 1.48 (1.26) | 0.86 |
Abbreviation: FNQ, Food Noise Questionnaire.
A summary of the responses to the FNQ and other validated questionnaires is provided in Table S3. Convergent validity was supported by strong positive correlations between the FNQ and the three‐item Frequency subscale of the Food Preoccupation Questionnaire (r = 0.79, p < 0.001) and the Preoccupation with Food subscale of the FCQ‐T (r = 0.87, p < 0.001; Table 4). In line with expectations, the correlations between the FNQ and the PHQ‐8 (r = 0.38, p < 0.001), the PSS (r = 0.37, p < 0.001), and the GAD‐7 (r = 0.31, p < 0.001) were low, which supports the discriminant validity of the FNQ (Table 4). Using Fisher's z‐transformation, the 95% CI for correlation coefficients for convergent validity were higher than those of all of the coefficients for discriminant validity (Table 4). The correlations were similar for the seven‐item version of the FNQ (data not shown). Together, these results support the construct validity of the FNQ.
TABLE 4.
Correlation of the total score of the final FNQ with the validated scales (n = 248)
| Measure | FNQ total score | ||
|---|---|---|---|
| r | p | 95% CI | |
| Frequency subscale of the Food Preoccupation Questionnaire | 0.79 | <0.001 | 0.735–0.831 |
| Preoccupation with Food subscale of the FCQ‐T | 0.87 | <0.001 | 0.830–0.894 |
| PHQ‐8 | 0.38 | <0.001 | 0.264–0.479 |
| PSS | 0.37 | <0.001 | 0.251–0.468 |
| GAD‐7 | 0.31 | <0.001 | 0.195–0.421 |
Abbreviations: FCQ‐T, Food Cravings Questionnaire‐Trait; FNQ, Food Noise Questionnaire; GAD‐7, General Anxiety Disorder‐7; PHQ‐8, 8‐item Patient Health Questionnaire; PSS, Perceived Stress Scale.
Differences by sample characteristics
Women had significantly higher FNQ scores than men (mean [SD] = 8.0 [5.4] and mean [SD] = 6.1 [5.1], respectively, p = 0.001; Table 5). Participants older than age 55 years and those who were retired had lower relative scores (mean [SD] = 6.0 [5.0] and mean [SD] = 6.0 [5.0], respectively, both p < 0.037). Those who were currently dieting to lose weight had higher scores than those who were not dieting to lose weight (mean [SD] = 9.1 [5.8] and mean [SD] = 6.7 [5.1], respectively, p < 0.001). Linear regression between BMI and FNQ scores showed that participants with higher BMI values tended to have higher FNQ scores (mean [SD] unstandardized slope = 0.1 [0.04], p = 0.011). There were no significant differences in mean FNQ scores across race and ethnicity, BMI category, household income, or education (all p > 0.05).
TABLE 5.
Differences in the total FNQ score by sample characteristics (n = 396)
| FNQ a score | |||||||
|---|---|---|---|---|---|---|---|
| N | Mean | SD | Min. | Median | Max. | p value | |
| Sex | |||||||
| Male | 130 | 6.06 | 5.06 | 0.00 | 5.00 | 20.00 | 0.001 |
| Female | 266 | 8.04 | 5.40 | 0.00 | 8.00 | 20.00 | |
| Race | |||||||
| White/non‐Hispanic | 286 | 7.55 | 5.41 | 0.00 | 7.00 | 20.00 | |
| Black or African American/non‐Hispanic | 48 | 6.40 | 5.05 | 0.00 | 6.00 | 20.00 | |
| Hispanic | 25 | 7.12 | 4.73 | 0.00 | 6.00 | 15.00 | 0.687 |
| Asian | 11 | 8.27 | 6.10 | 0.00 | 10.00 | 20.00 | |
| Other or prefer not to answer | 26 | 7.31 | 5.90 | 0.00 | 6.50 | 20.00 | |
| Ethnicity | |||||||
| Hispanic | 25 | 7.12 | 4.73 | 0.00 | 6.00 | 15.00 | |
| Non‐Hispanic | 361 | 7.34 | 5.37 | 0.00 | 6.00 | 20.00 | 0.38 |
| Prefer not to answer | 10 | 9.70 | 6.75 | 0.00 | 8.50 | 20.00 | |
| BMI, kg/m2 | |||||||
| 18.5–24.9 | 72 | 6.60 | 5.43 | 0.00 | 5.00 | 20.00 | |
| 25–29.9 | 122 | 7.14 | 5.01 | 0.00 | 7.00 | 20.00 | |
| 30–39.9 | 166 | 7.51 | 5.32 | 0.00 | 7.00 | 20.00 | 0.100 |
| >40 | 36 | 9.25 | 6.34 | 0.00 | 9.50 | 20.00 | |
| Dieting to lose weight? | |||||||
| Yes | 113 | 9.05 | 5.75 | 0.00 | 9.00 | 20.00 | <0.001 |
| No | 283 | 6.72 | 5.07 | 0.00 | 6.00 | 20.00 | |
| Age category, y | |||||||
| 18–34 | 45 | 8.49 | 5.25 | 0.00 | 10.00 | 20.00 | |
| 35–54 | 195 | 8.26 | 5.46 | 0.00 | 8.00 | 20.00 | 0.0001 |
| >55 | 156 | 5.98 b | 5.01 | 0.00 | 5.00 | 20.00 | |
| Education | |||||||
| Some high school | 11 | 9.18 | 5.76 | 0.00 | 10.00 | 20.00 | |
| High school diploma/GED | 81 | 7.43 | 5.06 | 0.00 | 7.00 | 20.00 | |
| Some college | 148 | 7.44 | 5.57 | 0.00 | 6.50 | 20.00 | 0.792 |
| Bachelor's degree | 106 | 7.34 | 5.29 | 0.00 | 6.00 | 20.00 | |
| Postgraduate degree | 50 | 6.88 | 5.45 | 0.00 | 6.50 | 20.00 | |
| Income | |||||||
| <$50,000 | 159 | 7.25 | 4.94 | 0.00 | 7.00 | 20.00 | |
| $50,000–$69,999 | 69 | 8.07 | 5.90 | 0.00 | 9.00 | 20.00 | |
| $70,000 and above | 161 | 7.16 | 5.58 | 0.00 | 6.00 | 20.00 | 0.545 |
| Prefer not to answer | 7 | 9.00 | 4.40 | 1.00 | 10.00 | 15.00 | |
| Employment status | |||||||
| Full‐time (40 h/wk) | 161 | 8.16 | 5.68 | 0.00 | 8.00 | 20.00 | |
| Other, please specify | 44 | 8.02 | 5.91 | 0.00 | 8.00 | 20.00 | |
| Part‐time | 53 | 7.32 | 4.51 | 0.00 | 6.00 | 20.00 | 0.036 |
| Retired | 89 | 6.02 c | 5.04 | 0.00 | 5.00 | 20.00 | |
| Unemployed | 49 | 6.84 | 4.85 | 0.00 | 7.00 | 18.00 | |
Abbreviations: FNQ, Food Noise Questionnaire; GED, General Educational Development; Max., maximum; Min., minimum.
FNQ scores can range from 0 to 20, with higher numbers indicating greater levels of food noise.
Participants older than age 55 years had significantly lower relative scores compared to other age groups.
Participants who were retired had significantly lower relative scores compared to other employment groups.
DISCUSSION
The findings indicate that the FNQ is a psychometrically reliable and valid tool for assessing food noise. The FNQ has a one‐factor structure with excellent internal consistency and high test–retest reliability. Construct validity was supported via tests of convergent and discriminant validity. The results support the reliability and validity of the FNQ and its use as a brief practical tool to measure food noise and to examine the demographic correlates of food noise. However, further research is needed to: 1) determine whether the food noise construct changes in response to various obesity treatments; 2) ensure the reliability of the construct by including longer or shorter time intervals between the questionnaire administration; and 3) identify whether the construct is more prominent among individuals who are seeking various weight loss treatments (e.g., obesity medications, bariatric surgery, lifestyle interventions). These further evaluations of the FNQ will enable researchers and clinicians to better understand the questionnaire's clinical utility.
Confirmatory factor analysis supported a one‐factor structure. Results showed that the fit indices favored a one‐factor structure, except for RMSEA; the values were higher than the cutoff. The low degrees of freedom observed can be attributed to this result. Specifically, Kenny et al. have noted that, even when the model is correctly specified, the RMSEA may exceed the cutoffs for models with small degrees of freedom [30]. The significant χ2 test can be explained by the sensitivity of this test to sample size. Large samples can result in significant χ2 values, even for well‐fitting models [23]. Additionally, a high item‐total correlation was observed for all items, which confirms that every item included in the FNQ assesses a single construct. These findings demonstrate the ability of the FNQ to measure the construct of food noise.
The FNQ had excellent internal consistency and high test–retest reliability, confirming the reliability of the FNQ. In line with the hypotheses, the total FNQ score was positively related to the score from the Frequency subscale of the Food Preoccupation Questionnaire [14] and the Preoccupation with Food subscale of the FCQ‐T [15]. According to the correlation between the FNQ and the Frequency subscale, individuals who reported higher levels of food noise also reported being frequently preoccupied with thoughts about food [14]. The size of the correlation coefficients indicates that the FNQ assesses a construct that is related to, but not redundant with, food preoccupation [14, 15]. Although the FNQ distinctly focuses on the food noise construct, the Preoccupation with Food subscale of the FCQ‐T assesses food cravings and reliving cravings by eating food [15]. Additionally, the Frequency subscale of the Food Preoccupation Questionnaire [14] and the FCQ‐T [15] assess food preoccupation, but they do not assess the extent to which the participants' food thoughts are intrusive and negatively impact their lives, which is assessed with the FNQ. Discriminant validity of the FNQ was confirmed by the lower correlations between the total scores of the PHQ‐8 [18], PSS [21], and GAD‐7 [22]. These findings demonstrate that the FNQ exhibits excellent psychometric properties and is an effective tool for measuring the construct of food noise.
Retired participants and participants older than age 55 years had lower FNQ scores. The FNQ scores were higher among women and those who were currently dieting to lose weight. These findings are consistent with those reported by Tapper and Pothos in a previous study in which female individuals and dieting individuals scored higher on the Frequency subscale of the Food Preoccupation Questionnaire [14]. The current study also found that FNQ scores increased with higher BMI values. These individual differences, however, should be viewed with caution. Given that the study's purpose was to develop a psychometrically valid measure, it might not adequately ascertain the extent to which the measure varies across demographic groups, the extent to which it has clinically significant sequelae, what would be considered clinically significant cut points, or its sensitivity to interventions. Now that a psychometrically valid measure of food noise exists, it is important to further validate the questionnaire in additional clinical and nonclinical settings with individuals seeking various obesity treatments.
The current study demonstrates the psychometric validity of the FNQ to measure participants' subjective experiences with food noise. According to research reports, GLP‐1 receptor agonists may affect the reward pathways in the brain [31, 32]; however, there are many potential mechanisms by which weight loss drugs can influence individuals' experiences with food noise. Therefore, in future studies, using the FNQ with other measures can help bridge the gap between the subjective experiences of food noise and the cognitive and physiological underpinnings of these experiences. For example, the FNQ can be used to examine the relationship between food noise and cognitive biases (e.g., attentional biases, memory biases), behavioral and appetitive tests (e.g., eating in the absence of hunger), and physiological measures related to food cue reactivity or the rewarding value of food (e.g., functional magnetic resonance imaging, heart rate, salivary response). This research could help identify the etiology of these constructs and elucidate how these phenomena are maintained through their interaction or their effect on food intake, avoidance, and related behaviors.
This study has several strengths. We developed a psychometrically sound measure of food noise by following rigorous scale development protocols and using validated measures for construct validity, and we provided strong empirical evidence of consistency, reliability, and validity. The sample was nationally representative in terms of race and ethnicity and BMI and had balanced distributions across age groups, and 82% of individuals reported living with overweight or obesity. The use of expert reviewers in refining the initial item pool facilitated face and content validity of the FNQ. The study also has some limitations. The FNQ was designed to be brief; however, it is possible that the five‐item questionnaire fails to capture other aspects of the food noise construct. Our test–retest reliability was assessed with a 7‐day time interval; therefore, to understand the short‐ and long‐term stability of the food noise construct, further investigation should be conducted using various time intervals. Although the sample is nationally representative, it is under‐indexed with respect to racial and ethnic groups that are most affected by obesity. The sample also included a greater proportion of female individuals than the targeted distribution, which may have influenced the results of the current study. Further examinations of the sensitivity and external validity of the FNQ were beyond the scope of the current study; therefore, future studies should test the extent to which FNQ scores change with weight loss and various weight management interventions.
CONCLUSION
In summary, the FNQ provides a valid, reliable, and concise self‐report measure that can be used to systematically and accurately quantify food noise and study the construct. The FNQ is available to use free of charge, with the hope of providing researchers and health care providers the opportunity to better understand food noise, its clinical and demographic correlates, and its sensitivity to various treatments.
AUTHOR CONTRIBUTIONS
Hanim E. Diktas, Monique M. LeBlanc, Gary D. Foster, Michelle I. Cardel, Rebecca Nathan, Danielle Shapiro, and Corby K. Martin designed the study and oversaw data acquisition and analysis. Stephanie L. Dickinson, Erin M. Ables, and Xiwei Chen analyzed data, and all authors participated in interpretation of the results. Hanim E. Diktas and Corby K. Martin wrote the first draft of the manuscript; all authors provided critical revisions for important intellectual content. The concept and design of the study were led by the principal investigator (Corby K. Martin) in collaboration with all authors. All authors read and approved the final manuscript.
FUNDING INFORMATION
The current study was funded by WW International, Inc. via a research contract awarded to Pennington Biomedical Research Center. Pennington Biomedical Research Center is supported by Nutrition Obesity Research Center (NORC) grant P30 DK072476 titled “Nutrition and Metabolic Health Through the Lifespan” sponsored by the National Institute for Diabetes and Digestive and Kidney Diseases (NIDDK) and by grant U54 GM104940 from the National Institute of General Medical Sciences, which funds the Louisiana Clinical and Translational Science Center.
CONFLICT OF INTEREST STATEMENT
Gary D. Foster, Rebecca Nathan, Danielle Shapiro, and Michelle I. Cardel are employees and/or shareholders of WW International, Inc. Stephanie L. Dickinson has done other statistical consulting with WW unrelated to this study. The other authors declared no conflicts of interest.
Supporting information
Data S1. Supporting Information.
ACKNOWLEDGMENTS
We thank the participants who took part in this study. We also thank the experts who have contributed to the development of the first version of the Food Noise Questionnaire, including Drs. Faith Anne Heeren, Hollie Raynor, Jamy Ard, John Apolzan, Kelly C. Allison, Rick Mattes, Robert F. Kushner, Thomas Wadden, and one additional expert who prefers to remain anonymous.
Diktas HE, Cardel MI, Foster GD, et al. Development and validation of the Food Noise Questionnaire. Obesity (Silver Spring). 2025;33(2):289‐297. doi: 10.1002/oby.24216
DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.
REFERENCES
- 1. Müller TD, Blüher M, Tschöp MH, DiMarchi RD. Anti‐obesity drug discovery: advances and challenges. Nat Rev Drug Discov. 2022;21:201‐223. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Blum D. People on drugs like Ozempic say their ‘food noise’ has disappeared. New York Times. June 21, 2023. Accessed October 10, 2024. https://www.nytimes.com/2023/06/21/well/eat/ozempic-food-noise.html
- 3. Cassata C. How Ozempic and Wegovy can help stop cravings and silence ‘food noise. Healthline website. Published June 28, 2023. Accessed October 10, 2024. https://www.healthline.com/health‐news/drugs‐like‐ozempic‐and‐wegovy‐cut‐cravings‐and‐turn‐down‐food‐noise
- 4. Brody B. What is food noise and how can you quiet it? WebMD website. Published September 21, 2023. Accessed October 10, 2024. https://www.webmd.com/obesity/features/what-is-food-noise
- 5. Hayashi D, Edwards C, Emond JA, et al. What is food noise? A conceptual model of food cue reactivity. Nutrients. 2023;15:4809. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Kuhn C. Patients say drugs like Ozempic help with ‘food noise.’ Here's what that means. PBS News website. Published September 25, 2023. Accessed October 10, 2024 2024. https://www.pbs.org/newshour/health/patients‐say‐drugs‐like‐ozempic‐help‐with‐food‐noise‐heres‐what‐that‐means
- 7. Boateng GO, Neilands TB, Frongillo EA, Melgar‐Quiñonez HR, Young SL. Best practices for developing and validating scales for health, social, and behavioral research: a primer. Front Public Health. 2018;6:149. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Olson K. An examination of questionnaire evaluation by expert reviewers. Field Methods. 2010;22:295‐318. [Google Scholar]
- 9. Comrey AL, Lee HB. A First Course in Factor Analysis. 2nd ed. Lawrence Erlbaum Associates, Inc; 1992. [Google Scholar]
- 10. Centers for Disease Control and Prevention . Adult obesity facts. Updated May 14, 2024. Accessed October 10, 2024. https://www.cdc.gov/obesity/adult-obesity-facts/
- 11. U.S. Census Bureau. QuickFacts: United States . Population estimates, July 1, 2023. Accessed October 10, 2024. https://www.census.gov/quickfacts/fact/table/US/PST045223
- 12. Collins LM. Research design and methods. In: Birren JE, ed. Encyclopedia of Gerontology. 2nd ed. Elsevier; 2007:433‐442. [Google Scholar]
- 13. Allen MJ, Yen WM. Introduction to Measurement Theory. Brooks/Cole Publishing Company; 1979. [Google Scholar]
- 14. Tapper K, Pothos EM. Development and validation of a Food Preoccupation Questionnaire. Eat Behav. 2010;11:45‐53. [DOI] [PubMed] [Google Scholar]
- 15. Cepeda‐Benito A, Gleaves DH, Williams TL, Erath SA. The development and validation of the State and Trait Food Cravings Questionnaires. Behav Ther. 2000;31:151‐173. [DOI] [PubMed] [Google Scholar]
- 16. Vander Wal JS, Johnston KA, Dhurandhar NV. Psychometric properties of the State and Trait Food Cravings Questionnaires among overweight and obese persons. Eat Behav. 2007;8:211‐223. [DOI] [PubMed] [Google Scholar]
- 17. Netemeyer RG, Bearden WO, Sharma S. Scaling Procedures. SAGE Publications Inc; 2003. [Google Scholar]
- 18. Kroenke K, Strine TW, Spitzer RL, Williams JBW, Berry JT, Mokdad AH. The PHQ‐8 as a measure of current depression in the general population. J Affect Disord. 2009;114:163‐173. [DOI] [PubMed] [Google Scholar]
- 19. Cheung RYM. Patient Health Questionnaire‐9 (PHQ‐9). In: Medvedev ON, Krägeloh CU, Siegert RJ, Singh NN, eds. Handbook of Assessment in Mindfulness Research. Springer International Publishing; 2024. doi: 10.1007/978-3-030-77644-2_63-2. [DOI] [Google Scholar]
- 20. de la Torre JA, Vilagut G, Ronaldson A, et al. Reliability and cross‐country equivalence of the 8‐item version of the Patient Health Questionnaire (PHQ‐8) for the assessment of depression: results from 27 countries in Europe. Lancet Reg Health Eur. 2023;31:100659. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Cohen S, Kamarck T, Mermelstein R. A global measure of perceived stress. J Health Soc Behav 1983;24:385–396. [PubMed] [Google Scholar]
- 22. Spitzer RL, Kroenke K, Williams JBW, Löwe B. A brief measure for assessing generalized anxiety disorder: the GAD‐7. Arch Intern Med. 2006;166:1092‐1097. [DOI] [PubMed] [Google Scholar]
- 23. Fayers EP, Hays R, eds. Assessing Quality of Life in Clinical Trials: Methods and Practice. 2nd ed. Oxford University Press; 2005. [Google Scholar]
- 24. Hu L, Bentler PM. Fit indices in covariance structure modeling: sensitivity to underparameterized model misspecification. Psychol Methods. 1998;3:424‐453. [Google Scholar]
- 25. Jöreskog KG, Sörbom D. LISREL 7: A Guide to the Program and Applications. SPSS; 1989. [Google Scholar]
- 26. Bentler PM. Comparative fit indexes in structural models. Psychol Bull. 1990;107:238‐246. [DOI] [PubMed] [Google Scholar]
- 27. Byrne BM. A Primer of LISREL: Basic Applications and Programming for Confirmatory Factor Analytic Models. Springer‐Verlag Publishing; 1989. [Google Scholar]
- 28. Flesch R. A new readability yardstick. J Appl Psychol. 1948;32:221‐233. [DOI] [PubMed] [Google Scholar]
- 29. Staffini A, Fujita K, Svensson AK, Chung U‐I, Svensson T. Statistical methods for item reduction in a representative lifestyle questionnaire: pilot questionnaire study. Interact J Med Res. 2022;11:e28692. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Kenny DA, Kaniskan B, McCoach DB. The performance of RMSEA in models with small degrees of freedom. Sociol Methods Res. 2015;44:486‐507. [Google Scholar]
- 31. Mietlicki‐Baase EG, Ortinski PI, Rupprecht LE, et al. The food intake‐suppressive effects of glucagon‐like peptide‐1 receptor signaling in the ventral tegmental area are mediated by AMPA/kainate receptors. Am J Physiol Endocrinol Metab. 2013;305:E1367‐E1374. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Alhadeff AL, Rupprecht LE, Hayes MR. GLP‐1 neurons in the nucleus of the solitary tract project directly to the ventral tegmental area and nucleus accumbens to control for food intake. Endocrinology. 2012;153:647‐658. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data S1. Supporting Information.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.