Acceptability and perceived harm of calorie labeling and other obesity policies: A cross‐sectional survey study of UK adults with eating disorders and other mental health conditions

Abstract Objective We assessed perceptions of recently proposed UK obesity policies (mandatory calorie labeling, banning of advertisements of unhealthy food and drinks online and before 9 pm on TV, and banning “buy one get one free” deals for unhealthy food and drinks) in people with an eating disorder (ED) and other mental health conditions. Method A total of 1273 participants with a self‐reported lifetime mental health condition (N = 583 with an ED) completed an online survey in September–November 2022. Multinomial logistic regression was used to examine support for and potential adverse effects of policies in participants with and without an ED. A qualitative analysis of the potential effects of the policy on current ED symptoms was also conducted to better understand how and why policies may be damaging or beneficial. Results Participants with an ED had a lower level of support for the implementation of the calorie labeling policy compared to those without an ED (43% vs. 58%). Half of the participants with an ED (55%) reported that labeling may worsen their ED symptoms. Qualitative data indicated perceived potential harm (e.g., a gateway to relapse, negative effects on mood) and perceived benefits (e.g., feeling informed and reassured) of calorie labeling in participants with an ED. No differences in support or perceived harms of the other two policies were observed between participants with versus without an ED. Discussion Future studies are warranted to explore the potential effects of calorie labeling and how to mitigate negative impacts on people with an ED. Public Significance This research is the first to assess the perceptions of UK obesity‐related policies in people with an ED and other mental health conditions. Participants with an ED (vs. without) were more likely to disagree with the government implementing the calorie labeling policy. These findings highlight the potentially harmful effects of calorie labeling in people with an ED and the need for future research to understand how to mitigate negative impacts.

score of four items that developed Factor 1 (the shape/weight overvaluation and body dissatisfaction subscales) to indicate ED symptomatology.This four-item measure was found to have very good internal consistency with Cronbach's Alpha = .93(see Table S2).

Table S2. Validity and reliability of EDE-Q7
Factor 2: Dietary restraint "On how many of the past 28 days ……"

Factor loadings
Cronbach's Alpha 1. "Have you been consciously trying to limit the amount of food you eat to influence your shape or weight?" .90 .90 2. "Have you attempted to avoid eating any foods which you like in order to influence your shape or weight?" .88 3. "Have you attempted to follow definite rules regarding your eating in order to influence your shape or weight; for example, a calorie limit, a set amount of food, or rules about what or when you should eat?" .88 Factor 1: Shape/weight overvaluation (items 4 and 5) Body dissatisfaction (items 6 and 7) "On how many over the past 28 days ……"

Factor loadings
Cronbach's Alpha 4. "Has your weight influenced how you think about (judge) yourself as a person?" .86 .93 5. "Has your shape influenced how you think about (judge) yourself as a person?" .87 6. "How dissatisfied have you felt about your weight?" .89 7. "How dissatisfied have you felt about your shape?" .89 A 4-item Patient Health Questionnaire (PHQ-4) EFA was also used to evaluate the validity and Cronbach's Alpha was used to determine the reliability of PHQ-4 (Fenn et al., 2020).Findings from EFA showed that only one factor was found with Eigenvalue > 1.0, and therefore, all the items can be aggregated together.All items also had high factor loadings (≥ .84).PHQ-4 also had good internal consistency (.88) (see Table S3).
Table S3.Validity and reliability of PHQ-4 PHQ-4 "Over the last 2 weeks, how often have you been bothered by the following problems?"

Robustness analysis: inverse probability weighting (IPW)
IPW approach was used to address missing values and potential selection bias due to some characteristics that may be associated with sample retention (Chesnaye et al., 2022;Mansournia & Altman, 2016).Based on the observed data, we found that increasing age and degree-level qualification completion were associated with having complete observations.Women and other genders were less likely to have complete observations compared to men.Using a logistic regression model, we estimated the probability of being retained in the analytical sample size (i.e., having complete vs. incomplete observations) based on sociodemographic characteristics associated with missingness (age, gender, educational level).We then calculated sample weights as the inverse probability of retention.We applied these weights in the multinomial logistic regression analysis to provide robust standard errors after taking into account differences in participants' characteristics associated with missingness.a The association remained significant at p < .01 when ED symptoms from EDE-Q7 (shape/weight overvaluation, body dissatisfaction) and mental health symptoms from PHQ-4 were also adjusted into the model (RRR = 0.60; 95% CI = 0.42, 0.88; p = .009;n = 902).Additional findings when IPW approach was used to address missing observations for the analyses of the perceptions of calorie labelling policy:

Additional tables for the results section
-We found no statistically significant differences in the perceptions of calorie labelling policy based on the status of current ED diagnosis (current vs. past diagnosis) when IPW approach was used, similar to the findings presented in Table S5 (full findings are not presented).
-We found no statistically significant associations between ED diagnoses (yes vs. no) and opinions on using and asking for a menu with calorie labelling when IPW approach was used, similar to the findings presented in Table S6 (full findings are not presented).
-We also used IPW approach in examining the associations between ED diagnoses (yes vs. no) and perceived negative effects of calorie labelling on the current ED and mental health symptoms (worse vs. neutral) (see Table 4 in the main document for the comparison).Findings from using IPW approach indicated that the associations between the following ED diagnoses and perceived negative effects of calorie labelling on current ED symptoms were significant at p < .05:anorexia nervosa (RRR = 2.20; 95% CI = 1.04, 4.63; p = .039),bulimia nervosa (RRR = 2.42, 95% CI = 1.24, 4.71; p = .01)(full findings are not presented).
Table S14.Differences in acceptability and perceptions of banning advertisements of unhealthy food and drinks online and before 9 pm on TV between participants who have and have not been diagnosed with an ED using inverse probability weighting approach

Table S4 .
Differences in acceptability and perceptions of mandatory calorie labelling on menus between participants who have and have not been diagnosed with an ED (models include additional adjustments for ED and mental health symptomology)

Table S5 .
Adjusted associations between the status of current ED and opinions on mandatory calorie labelling policy in participants who have been diagnosed with an ED *p < .05;**p< .01;***p< .001RRR= relative risk ratio; CI = confidence interval; ref = reference group; past ED diagnosis = "Past ED and no current symptoms" + "Past ED and some lingering symptoms"; current ED diagnosis = "Currently ED and experience symptoms"Multinomial logistic regression models were developed for each item of acceptability and perceptions of the policy, adjusting for age, gender, ethnicity, education, tertiles of equivalised household income, and BMI category.

Table S6 .
Adjusted associations between diagnoses of ED and other mental health conditions and opinions on the use of calorie information in participants who have been diagnosed with an ED *p < .05;** p <.01; *** p <.001 RRR = relative risk ratio; CI = confidence interval; ref = reference group Multinomial logistic regression models were developed for each item of acceptability and perceptions of the policy, adjusting for age, gender, ethnicity, education, tertiles of equivalised household income, and BMI category.

Table S7 .
Adjusted associations between diagnoses of ED and other mental health conditions and perceived effect of calorie labelling policy on the current symptoms among participants who have been diagnosed with an ED (models include additional adjustments for ED and mental health symptomology) *p < .05;**p < .01;***p < .001RRR = relative risk ratio; CI = confidence interval; ref = reference group Multinomial logistic regression models were developed for each item of acceptability and perceptions of the policy, adjusting for age, gender, ethnicity, education, tertiles of equivalised household income, BMI category, ED symptoms from EDE-Q7 (shape/weight overvaluation, body dissatisfaction), and mental health symptoms from PHQ-4.

Table S8 .
Acceptability and perceptions of banning advertisements of unhealthy food and drinks online and before 9 pm on TV in participants who have and have not been diagnosed with an ED marketing and advertising of unhealthy foods and drinks (i.e., high fat, salt and/or sugar products) online and before 9 pm on TV will not provide benefits to children's health.
A ban on marketing and advertising of unhealthy foods and drinks (i.e., high fat, salt and/or sugar products) online and before 9 pm on TV will make my eating disorder symptoms…* *This item was only administered to participants who have been diagnosed with an ED n = number of participants; % = percentage; ED diagnosis= participants who have been diagnosed with an ED; No ED diagnosis= participants who have not been diagnosed with an ED

Table S9 .
Differences in acceptability and perceptions of banning advertisements of unhealthy food and drinks online and before 9 pm on TV between participants who have and have not been diagnosed with an ED *p < .05;**p < .01;***p < .001RRR = relative risk ratio; CI = confidence interval; ref = reference group; ED diagnosis = participants who have been diagnosed with an ED; No ED diagnosis = participants who have not been diagnosed with an ED Multinomial logistic regression models were developed for each item of acceptability and perceptions of the policy, adjusting for age, gender, ethnicity, education, tertiles of equivalised household income, and BMI category.

Table S10 .
Acceptability and perceptions of banning "buy one get one free" deals for unhealthy food and drinks in participants who have and have not been diagnosed with an ED A ban on "buy one get one free" deals for unhealthy foods and drinks (i.e., high fat, salt and/or sugar products) will make my eating disorder symptoms…* *This item was only administered to participants who have been diagnosed with an ED n = number of participants; % = percentage; ED diagnosis= participants who have been diagnosed with an ED; No ED diagnosis= participants who have not been diagnosed with an ED

Table S11 .
Differences in acceptability and perceptions of banning "buy one get one free" deals for unhealthy food and drinks between participants who have and have not been diagnosed with an ED *p < .05;**p < .01;***p < .001RRR = relative risk ratio; CI = confidence interval; ref = reference group; ED diagnosis = participants who have been diagnosed with an ED; No ED diagnosis = participants who have not been diagnosed with an ED Multinomial logistic regression models were developed for each item of acceptability and perceptions of the policy, adjusting for age, gender, ethnicity, education, tertiles of equivalised household income, and BMI category.

Table S12 .
Analyses examining differences in acceptability and perceptions of mandatory calorie labelling on menus between participants who have and have not been diagnosed with an ED using inverse probability weighting approach

Table S13 .
Differences in acceptability and perceptions of mandatory calorie labelling on menus between participants who have and have not been diagnosed with an ED (models include additional adjustments for ED and mental health symptomology) using inverse probability weighting approach marketing and advertising of unhealthy foods and drinks (i.e., high fat, salt and/or sugar products) online and before 9 pm on TV will make my other mental health symptoms(n = 873; ref = No ED diagnosis)p < .05;**p < .01;***p < .001RRR = relative risk ratio; CI = confidence interval; ref = reference group; ED diagnosis = participants who have been diagnosed with an ED; No ED diagnosis = participants who have not been diagnosed with an ED Multinomial logistic regression models were developed for each item of acceptability and perceptions of the policy, adjusting for age, gender, ethnicity, education, tertiles of equivalised household income, and BMI category. *

Table S15 .
Differences in acceptability and perceptions of banning "buy one get one free" deals for unhealthy food and drinks between participants who have and have not been diagnosed with an ED using inverse probability weighting approach Buy one get one free" deals for unhealthyfoods and drinks (i.e., high fat, salt and/orsugar products) should be banned (n = 915; ref = No ED diagnosis) "buy one get one free" deals for unhealthy foods and drinks (i.e., high fat, salt and/or sugar products) would be helpful to prevent from buying more unhealthy foods and drinks (n = 915 ref = No ED diagnosis) "buy one get one free" deals for unhealthy foods and drinks (i.e., high fat, salt and/or sugar products) will not provide benefits to people's health (n = 914; ref = No ED diagnosis) "buy one get one free" deals for unhealthy foods and drinks (i.e., high fat, salt and/or sugar products) will make my other mental health symptoms (n = 873; ref = No ED diagnosis) p < .05;**p < .01;***p < .001RRR = relative risk ratio; CI = confidence interval; ref = reference group; ED diagnosis = participants who have been diagnosed with an ED; No ED diagnosis = participants who have not been diagnosed with an ED Multinomial logistic regression models were developed for each item of acceptability and perceptions of the policy, adjusting for age, gender, ethnicity, education, tertiles of equivalised household income, and BMI category. *