Polish Adaptation and Validation of the Intuitive (IES-2) and Mindful (MES) Eating Scales—The Relationship of the Concepts with Healthy and Unhealthy Food Intake (a Cross-Sectional Study)

Intuitive (IE) and mindful (ME) eating share internally focused eating, yet previous studies have shown that these concepts are not strongly correlated, which suggests that they might be differently related to food intake. The study aimed to adapt the original Intuitive (IES-2) and Mindful (MES) Eating Scales to the Polish language, to test their psychometric parameters and, further, to examine associations of IE and ME with an intake of selected food groups, i.e., healthy foods (fresh and processed vegetables, fresh fruit) and unhealthy foods (sweets, salty snacks). A cross-sectional study was conducted in 2020 in a group of 1000 Polish adults (500 women and 500 men) aged 18–65 (mean age = 41.3 ± 13.6 years). The factor structure was assessed with exploratory (EFA) and confirmatory (CFA) factor analysis as well as structural equation modeling (SEM). Measurement invariance across gender was assessed with multiple-group analysis. Internal consistency and discriminant validity of the two scales was tested. Spearman’s correlation coefficient was used to examine the correlation between IES-2 and MES subscales with food intake. A 4-factor, 16-item structure was confirmed for IES-2, while EFA and CFA revealed a 3-factor, 17-item structure of MES. Both scales demonstrated adequate internal consistency and discriminant validity. Full metric and partial scalar invariance were found for IES-2, while MES proved partial invariances. “Awareness” (MES) and “Body–Food Choice Congruence” (IES-2) positively correlated with intake of healthy foods and negatively with the intake of unhealthy ones. “Eating For Physical Rather Than Emotional Reasons” (IES-2) and “Act with awareness” (MES) favored lower intake of unhealthy foods, whereas “Unconditional Permission to Eat” and “Reliance on Hunger and Satiety Cues” (IES-2) showed an inverse relationship. A greater score in “Acceptance” (MES) was conducive to lower intake of all foods except sweets. The results confirmed that adapted versions of the IES-2 and MES are valid and reliable measures to assess IE and ME among Polish adults. Different IE and ME domains may similarly explain intake of healthy and unhealthy foods, yet within a single eating style, individual domains might have the opposite effect. Future studies should confirm our findings with the inclusion of mediating factors, such as other eating styles, childhood experiences, dieting, etc.


Introduction
Intuitive and mindful eating are the two most commonly listed non-diet practices focused on the elimination of eating in response to external cues [1]. With the growing popularity of alternative weight management methods and also among health practitioners [2,3], it is of interest which approach would be most beneficial for maintaining health [1,4,5].
Intuitive eating (IE) is described as a process integrating food, mind, and body, which involves rejection of diet mentality [6,7]. Ten major IE principles were developed in 1995 assess association of IE and ME with the intake of selected food groups, i.e., vegetables (fresh and processed), fresh fruit, sweets, and salty snacks, in women and men. We hypothesize that the original structure of the IES-2 and MES would not be replicated due to the cultural differences, yet psychometric parameters of the obtained models will be satisfactory. Moreover, our hypothesis is that IE and ME domains would not be strongly correlated and that they will be differently correlated with food intake, including differences observed between a group of women and men.

Study Design and Sample Collection
Data were collected in February 2020 through a cross-sectional survey. The study sample was recruited by a professional market research agency from an e-panel consisting of around 60,000 registered individuals. CAWI (Computer-Assisted Web Interview) technique was used to collect survey data. To ensure sample representativeness, quota controls, such as age, place of residence, and region, were set. Only participants aged 18-65 years were included in the study. Participance in the study was voluntary. Informed consent was obtained from all the study participants before the study. The final study sample included 1000 participants (500 women and 500 men). Data confidentiality as well as anonymity were assured. The study was approved by the Ethics Committee of the Institute of Human Nutrition Sciences, Warsaw University of Life Sciences, in Poland (Resolution No. 02/2020).

Instruments: Intuitive Eating Scale-2 (IES-2) and Mindful Eating Scale (MES)
Intuitive eating was assessed with the Intuitive Eating Scale-2 (IES-2) [9]. This 23-item measure consists of 4 subscales: (1) Unconditional Permission to Eat (6 items, e.g., "If I am craving a certain food, I allow myself to have it."); (2) Eating for Physical Rather Than Emotional Reasons (8 items, e.g., "I am able to cope with my negative emotions (e.g., anxiety, sadness) without turning to food for comfort."); (3) Reliance on Hunger and Satiety Cues (6 items, e.g., "I rely on my hunger signals to tell me when to eat."); and (4) Body-Food Choice Congruence (3 items, e.g., "I mostly eat foods that give my body energy and stamina."). In the current study, original coding and scoring procedure was used. Participants were asked to rate each item from strongly disagree (1) to strongly agree (5). A mean score was calculated for each subscale (range 1-5), with higher scores reflecting higher level of intuitive eating.
Polish adaption of the IES-2 and MES was conducted by two independent translators. The process included several stages: initial translation into Polish, synthesis of the two versions, back-translation into English, corrections, preparation of the prefinal version, pretesting in a group of 49 students, and further agreement on the final version (Appendices A and B) [37].

Food Intake
A food frequency questionnaire (KomPAN ® ) [38] was used to measure the frequency of the consumption of the selected food groups in the last three months: vegetables (fresh and processed, separately), fresh fruit, sweets, and salty snacks. The frequency was evaluated in 6 following categories: never (1), 1-3 times a month (2), once a week (3), a few times a week (4), once a day (5), and a few times in a day (6). Those values were converted to reflect the daily frequency of intake, ranging from 0 to 2 times/day [39]. Moreover, participants declared how many portions of products they eat daily given that 1 portion of vegetables and fruit (both fresh and processed) equals 100 g, and 1 portion of sweets and salty snacks are 50 g. Exemplary portion sizes were added to each question. To obtain the final value reflecting the daily intake of each food group, the daily frequency of intake was multiplied by portions consumed per day. Daily food intake was categorized into terciles, with the 1st tercile indicating "low intake" and 3rd tercile "high intake".

Statistical Analysis
Descriptive statistics were used to present sociodemographic characteristics. The diversity of socio-demographics between women and men was tested with the independence χ 2 test.
Exploratory factor analysis (EFA) with varimax rotation with Kaiser normalization was conducted to confirm the factorial structure of the IES-2 and MES. Certain criteria were set to determine final factors: eigenvalue ≥ 1.0, a scree plot test, interpretability of the solution, and factor loadings of at least 0.50. Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy and Bartlett's test of sphericity confirmed data factorability [40].
The fit of the factorial structure identified during EFA was tested by running confirmatory factor analysis (CFA). The following model fit indices were assessed: the chi-square fit statistics/degree of freedom (χ 2 /df), Tucker-Lewis index (TLI), comparative fix index (CFI), root mean square error of approximation (RMSEA), and root mean square residual (RMR). Acceptable values of mentioned parameters were: χ 2 /df below 2 or 3; TLI ≥ 0.95; CFI ≥ 0.95; RMSEA < 0.06; and the smallest RMR possible [41]. Models were modified until they revealed satisfactory fit parameters by correlating error variances and eliminating items with low loading values and/or high standardized residual covariances (>4) with multiple items [42].
Configural, metric, and scalar invariance of the final models of IES-2 and MES were tested across gender [43]. The chi-square difference test was selected to examine changes in the fit among models. p-Value above 0.05 indicated non-significant changes [43]. If the p-value was below 0.05, items that were causing a significant decrease of the model fit were allowed to be freely estimated across gender groups to achieve partial measurement invariance [43].
The internal consistency of items within each identified factor was tested using Cronbach's alpha, with values higher than 0.70 considered acceptable.
Mean factor scores for the IES-2 and MES in the total sample and separately in women and men were presented with descriptive statistics. Normality of the distribution was checked by conducting The Shapiro-Wilk test. The Mann-Whitney U Test was applied to compare differences between women and men in factor scores. p-Value lower than 0.05 was considered significant.
Spearman's correlation was used to measure the association between intuitive and mindful eating components as well as their relationship with the intake of selected food group products.
Discriminant validity was assessed by running the Mann-Whitney U test. Mean scores for IES-2 and MES subscales were compared between two groups identified based on intake of favorable (fresh and processed vegetables and fresh fruit) and unfavorable (sweets and salty snacks) foods, i.e., low intake, 1st tercile; high intake, 3rd tercile.
The analyses were carried out using IBM SPSS Statistics for Windows, version 26.0 (IBM Corp, Armonk, NY, USA) and AMOS graphics version 27.0.

Characteristics of the Study Sample
Sociodemographic characteristics of the study sample are presented in Table 1. The study group included 1000 individuals (equally numbered groups of women and men) aged 18-65 years, with a mean age of 41.3 years (±13.6 standard deviation (SD)). There were no significant differences among gender groups according to age, education, and place of residence.

Factor Structure of the IES-2
EFA revealed a four-factor structure of the IES-2. The total variance explained was 60.29%. Three items (Item 12, 13, and 15) had a factor loading below 0.5; thus, they were eliminated from further analysis ( Table 2). Confirmatory Factor Analysis (CFA) was used to test the fit of the model with four correlated latent variables identified during EFA ( Table 3). As the initial model proved unsatisfactory parameters (Model a), several modifications were introduced. Firstly, Item 14 was eliminated due to the low loading value (0.66) (Model b). Then, based on modification indices, error variances of the items within the same factor were allowed to correlate (Items 8 and 23, 1 and 9, 1 and 4, 1 and 20, 9 and 19, 4 and 9, 19 and 20, 4 and 20; Model c). Item 4 was further removed due to the low loading value (0.67; Model d). Finally, Item 9 was eliminated, as it had high residual covariances (Model e). The final model consisted of 16 items within four factors named as follows: Factor 1-"Reliance on Hunger and Satiety Cues" (RHSC, Items 6-8 and 21-23, Cronbach's alpha = 0.888); Factor 2-"Eating For Physical Rather Than Emotional Reasons" (EPR, Items 2, 5, 10, and 11, Cronbach's alpha = 0.859); Factor 3-"Body-Food Choice Congruence" (B-FCC, Items 1, 19, and 20, Cronbach's alpha = 0.763); Factor 4-"Unconditional Permission to Eat" (UPE, Items 3, 16, and 17, Cronbach's alpha = 0.747).

Multi-Group Analysis
Measurement invariance across gender was tested by comparing fit measures of created models (Table 4). Metric invariance was supported (p > 0.05), whereas scalar invariance was not confirmed (p < 0.05). Three items were found to significantly decrease model fit (Item 7, p = 0.020; Item 1, p < 0.001; and Item 16, p = 0.028); thus, they were allowed to be freely estimated by releasing the constraints across gender groups. As a result, partial scalar invariance was achieved (Adj. Model 2).
Mean scores for IES-2 subscales were calculated (Table 5). Based on EFA results, originally reverse-scored Item 1 turned out to have negative factor loading; thus, reversescoring of this item was no longer needed. Females were found to present a higher level of "Body-Food Choice Congruence", whereas men displayed greater intensity of "Eating For Physical Rather Than Emotional Reasons". Two remaining subscales did not significantly differ among women and men.

Multi-Group Analysis
Fit measures of created models were compared to test measurement invariance across gender (Table 8). Metric non-invariance was found (p < 0.05); thus, the constraints for sources of non-invariance (Item 20, p = 0.002; Item 13, p = 0.014) were released, and the model was retested, resulting in partial metric invariance (Adj. Model 1). Scalar invariance was not confirmed either (p < 0.05); therefore, the constraints for two items (Item 11, p = 0.004; Item 18, p = 0.010), which were found to decrease model fit, were released, resulting in partial scalar invariance (Adj. Model 2). Mean scores were calculated for each identified MES subscale (Table 9). Significant difference between women and men was noted only for "Awareness" subscale.

Relationship between Intuitive and Mindful Eating and Their Association with Food Intake
The relationship between IES-2 and MES factors was tested (Table 10). Moderate correlations were found between "Eating For Physical Rather Than Emotional Reasons" and two MES subscales-"Act with awareness" and "Acceptance" (r = 0.487 and r = 0.401, respectively).
The discriminant capability of IES-2 and MES was presented in Table 11. Participants characterized by high intake (3rd tercile) of fresh and processed vegetables and fresh fruit scored higher in "Awareness" and "Body-Food Choice Congruence" subscales. Those characterized by low intake of these products (1st tercile) scored higher in "Acceptance" and "Eating For Physical Rather Than Emotional Reasons". Higher scores for "Act with awareness", "Acceptance", "Eating For Physical Rather Than Emotional Reasons", and "Body-Food Choice Congruence" favored lower intake of sweets and salty snacks, whereas participants from the 3rd tercile of sweets and salty snacks intake scored higher in "Reliance on Hunger and Satiety Cues" and "Unconditional Permission to Eat".  IES-2 and MES factors were correlated with the intake of selected food groups in grams (Table 12). Awa, similar to B-FCC, positively correlated with intake of both fresh and processed vegetables and fresh fruit in the total group as well as in women and men separately. Positive correlations were also observed for RHSC and intake of fresh fruit (except for men) as well as sweets, which were also found to correlate with UPE yet in a greater manner among women than in men. It was noted that the higher the score in UPE, the greater the intake of salty snacks in women. Negative correlations were found between ActAwa and EPR and intake of sweets and salty snacks. Awa also negatively correlated with intake of salty snacks in the total group and both women and men. The higher the Acc score, the lower was the intake of salty snacks and fresh vegetables in women and men and fresh fruit in men. A greater score in B-FCC was linked with a lower intake of sweets in women.

Discussion
The original version of the Intuitive Eating Scale-2 [9] was previously adapted and validated in diverse populations [14,[44][45][46][47][48][49][50][51][52][53]. Most studies confirmed a four-factor structure of the IES-2 [44,45,[47][48][49][50][51]53]; yet, it should be noted that women, especially young or middle-aged, constituted the majority of the study samples [44,45,47,[49][50][51]. A few studies included a similar number of women and men within a broader age range [46][47][48]53], yet only two of them confirmed the original factorial structure of the IES-2 [48,53], similar to our results obtained from a representative study sample consisting of adults aged 18-65. Those incoherent results suggest that the IES-2 structure may vary in different settings and cannot be replicated in any sample without previous testing. All items within our final model loaded into the same factors as in the original IES-2 except Item 1, originally reverse-scored and loaded into UPE subscale, i.e., "I try to avoid certain foods high in fat, carbohydrates, or calories". In the current study, it loaded into the B-FCC subscale, and reverse-scoring was revoked based on EFA results. B-FCC is described as an ability to match food choices with one's body needs, i.e., eating mostly nutritious foods that increase energy levels, etc. [9]. Limiting intake of foods that worsen body functioning as a conscious, voluntary choice yet not in a form of imposed restrictions seems to be in line with B-FCC.
Despite being used in several studies [11,14,[34][35][36], Mindful Eating Scale [29] was not previously adequately tested in different cultures, languages, and countries. In our study, the original six-factor MES structure was not confirmed, and several items have been removed, as they negatively affected model fit indices. Our results suggest that more studies with MES are needed to confirm its structure, including testing correlations between MES and other scales used to measure ME [28,[30][31][32][33] to ensure that MES is a reliable and valid tool to assess ME.
According to the research, intuitive and mindful eating may be associated with a greater ability of food intake regulation [4,12,16,17]. However, the outcomes of the previous studies assessing how intuitive and mindful eating are linked with certain dietary habits are inconsistent, which confirms the need for further investigation [1,[18][19][20][21][22][23][24][25][26][27]. Some studies suggest possible beneficial correlations between selected components of those eating styles and eating behaviors, including greater intake of favorable foods, such as fruit and vegeta-bles, and a lower intake of energy-dense food [18,[20][21][22][23]54]. This was partially confirmed by our results, which primarily concern the positive relationship between Awa (MES) and B-FCC (IES-2) scores and consumption of fresh and processed vegetables and fresh fruit. Making conscious food choices, for both men and women, including appearance, smell, and taste of food but also the adjustment of food intake to the body (its good functioning, energy, condition) accompanied by lower intake of products high in fat, carbohydrates, and calories, promotes consumption of favorable foods.
Although in previous studies, intuitive eating subscales (except for UPE) were generally associated with healthier food choices and greater diet quality [18,20,21], in our study, it was confirmed only for B-FCC and intake of fresh fruit and fresh and processed vegetables. In addition, a positive correlation between intake of fresh fruit and reliance on hunger and body signals (RHSC) was observed in the female group. On the other hand, lack of emotional eating (EPR) was associated with a lower intake of sweets and salty snacks in both men and women. Thus, the positive association with higher dietary quality was also confirmed.
As in other studies, a higher UPE score as well as RHSC was not associated with higher diet quality, as it correlated positively with the intake of sweets. Previous studies on intuitive eating have confirmed the relationship of UPE with a greater intake of sweets and salty snacks [18,20,21]. In our study, only women with a higher score of unconditional permission to eat showed a higher intake of salty snacks. Furthermore, a lower intake of fruit and vegetables was not confirmed [18,20,21], and even a weak positive correlation with fresh fruit consumption has been shown. The positive correlations between RHSC, UPE, and intake of sweets might show that intuitive eating, especially in the initial phase, can be challenging to implement, resulting in less favorable eating behaviors, as this approach rejects restrictions in quality and quantity of food [12]. RHSC is characterized by eating in response to internal hunger and satiety cues [7]. Despite being an innate ability, RHSC may be disrupted by childhood food experiences associated with parental feeding practices that cause children to consume more food than the body needs [55,56]. Dieting, disordered eating, or eating disorders appearing in the past can also alter the sensation of hunger and satiety [7,57]. For example, higher intake of unfavorable foods as a result of unconditional permission to eat what is desired (UPE) might be only a temporary effect observed among individuals previously engaging in rigid dietary control and restrictions, known as a risk factor for excessive consumption or eating for reasons unrelated to physical hunger [58,59]. Our results may be due to the influence of previous individuals' experiences, yet such factors were not included in our study. It is suggested to incorporate them as mediators into future studies on the relationship between eating style, eating behavior, and food intake.
"Awareness" (Awa), which concerns physical sensations that occur while eating, such as food aroma, textures, smell, and how it looks [29], positively correlated with consumption of fresh and processed vegetables and fresh fruit. In contrast, greater ActAwa and Acc correlated with lower consumption of unfavorable foods (sweets and salty snacks), which may improve diet quality. Similar observations, especially in comparison with ActAwa, were noted for the EPR factor (IES-2). Both ActAwa and Acc were moderately positively correlated with EPR (IES-2), which may explain these findings. Nevertheless, Acc also correlated with lower consumption of fresh and processed vegetables. Thus, this element of mindful eating may contribute to a lower amount of food consumed in general.
Several differences among women and men were found, which proves that studies on IE and ME and their association with food intake should examine this relationship separately for those two groups. Among women, ActAwa and RHSC correlated positively with fresh fruit intake and UPE with salty snacks intake, whereas a negative correlation was found for B-FCC and sweets. On the other hand, in men, negative correlations were observed for fresh vegetables and EPR as well as for fresh fruit and Acc, while Awa was positively correlated with sweets intake. Significant differences between women and men in EPR, B-FCC, and Awa scores (Tables 5 and 9) may partially explain some of our observations. Another factor that could have affected those results are differences in dietary intake among women and men, i.e., lower/greater intake of certain foods [60][61][62]. Additionally, the above-mentioned negative experiences from the past might have had mediating role [7,[55][56][57].
According to the hypothesis, original structure of the IES-2 and MES was not replicated, which can both result from cultural differences and the use of representative study sample; nevertheless, final models proved satisfactory psychometric parameters. IE and ME domains were not strongly correlated, as expected, which may explain several differences observed in their relationship with food intake. As intended, correlations between IE and ME domains with food intake varied between women and men.
The current study as well as others conducted with the use of validated versions of IES-2 and MES will be useful for better understanding of the factors, such as feelings, food motives, and thoughts about food, in explaining food choices. In addition, their results will provide a better understanding of alternative weight control methods, which simultaneously positively influence food intake, i.e., higher consumption of favorable and lower consumption of unfavorable foods. Observations on gender differences in IE, ME, and food choices might be used in developing strategies addressed directly to women and men, aimed at improving diet quality, health, and food beliefs.

Strengths and Limitations
Adaptation and validation of the IES-2 and MES were conducted in a sample representative for the Polish population, which can be pointed as a study strength. Moreover, this is the first study to assess simultaneously IE and ME and their correlation with food intake among Polish adults. Nevertheless, a few limitations should be mentioned. Convergent validity of the IES-2 and MES as well as their test-retest reliability were not tested in our study. Moreover, the causality of associations cannot be determined due to the cross-sectional design of the study. Correlations between IE, ME, and food intake were examined only with the use of selected food groups. As the study was conducted in February, and questions about food intake related to the last 3 months of the consumption, the aspect of seasonal changes in dietary intake was not included.

Conclusions
Our study supported a 4-factor, 16-item structure of the Intuitive Eating Scale-2 (IES-2), while a 3-factor, 17-item solution was revealed for the Mindful Eating Scale (MES). A valid Polish version of IES-2 and MES might be useful in further investigation of factors influencing eating behaviors in this population. Different IE and ME domains may similarly explain intake of healthy and unhealthy foods yet within a single eating style individual domains might have the opposite effect. Nonetheless, future studies should confirm our findings on IE, ME, and food intake relationships in women and me, also with the use of other scales to measure ME. It is also suggested that further research would include mediating the role of factors, such as other eating styles, childhood experiences, dieting, or eating disorder episodes, on the relationship between IE, ME, and healthy or unhealthy food intake.

Informed Consent Statement:
Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.