A multifaceted approach to assessment of mentalization: the mentalization profile in patients with eating disorders

Abstract

Purpose: The existing literature, however sparse, suggests an association between eating disorders (ED) and mentalization ability. The aim of this study was to investigate the mentalization profile (MP) in patients with ED. It was hypothesized that patients with ED would have a lower degree of mentalization ability compared to healthy controls (HC).

Materials and methods: The study is based on a cross-sectional survey on a sample of patients diagnosed with ED compared to a HC group.

Results and conclusion: A total of 88 participants, distributed between patients with ED (N = 30) and HC (N = 58) were included.

Results: The study results show statistically significant differences between patients with ED and HC. Thus, patients with an ED scored significantly higher on Reflective Functioning Questionnaire (uncertainty about mental states) (RFQ-U) (mean difference: 0.31, p = 0.048) and Toronto Alexithymia Scale (TAS-20) (mean difference: 0.44, p = 0.019) compared to the HC. Furthermore, the results indicated that patients with Bulimia Nervosa (BN) may have a lower ability to mentalize about oneself as well as a generally lower ability to mentalize across different dimensions of the mentalization profile as the BN group scored significantly higher on RFQ-U (mean difference: 0.71, p = 0.023) and TAS-20 (mean difference: 0.89, p = 0.006) compared to the Anorexia Nervosa (AN) group. Further research should be conducted to gain knowledge on the mentalization ability in patients with an ED.

Keywords:

• Mentalization profile
• reflective functioning
• eating disorders
• alexithymia
• RFQ

Background

Eating disorders (ED) are characterized by disturbed eating behavior resulting in impaired physical health and psychosocial functioning. There are several subtypes of ED including anorexia nervosa (AN) and bulimia nervosa (BN). AN is characterized by fear of gaining weight, and a disturbed perception of body shape. A restrictive eating pattern is present in AN resulting in significant weight loss and/or being underweight. BN is characterized by binge eating with subsequent compensatory behaviors, e.g. self-induced vomiting or misuse of laxatives. BN often includes severe psychological symptoms, and self-evaluation is over-dependent on body shape and weight [1]. The existing literature suggests an association between ED and the mentalization ability. Mentalization can be defined as the ability to understand and interpret one’s own and others’ behavior based on the underlying mental states such as feelings and thoughts. Factors in early life such as insecure attachment and failed parenting practices are considered to predispose for impaired mentalization ability and psychopathology later in life such as, but not limited to, ED [2–4]. Impaired mentalization may make the individual vulnerable to social and cultural factors, which are also known as predisposing factors to ED [5–7]. The ability to mentalize is multifunctional, and the different dimensions of mentalization can be described as four sets of functional polarities that are highly contextual: (1) Automatic – Controlled, (2) Internally focused – Externally focused, (3) Self-oriented – Other-oriented and (4) Cognitive process – Affective process. The term MP can advantageously be used to emphasize the ability to mentalize as multidimensional. The MP may be inhibited in some dimensions but not necessarily in all [5,8,9]. As the MP is multidimensional, parts of the ability to mentalize can be described through already known psychological concepts such as Theory of Mind (ToM) and alexithymia. Such concepts should not be conflated with the MP but only as components hereof. ToM describes the ability of the individual to attribute mental states to others, and as such this concept only describes the dimension of the MP that concerns mentalization about others. Alexithymia is the inability to understand, identify and describe own emotional states. A high degree of alexithymia means larger difficulties in describing own feelings and is an expression of inhibited ability to mentalize about the self. The MP is very complex as its many dimensions can be investigated and described through many psychological concepts [6,8,10]. This emphasizes the need to include several different tests that investigate different dimensions and contexts of the ability to mentalize in order to describe the MP.

The existing literature on the MP in patients with ED is limited. A systematic review and meta-analysis by [11] found the MP in patients with an ED to be characterized by more inhibited mentalization about oneself compared to HC, while the ability to mentalize about others seemed comparable to HC. While the study showed significantly lower ability to mentalize about the self in AN when compared to HC, it was also concluded that patients with AN had a tendency to lower mentalization ability when compared to patients with BN [11]. Further research has found impaired mentalization in patients with ED [12–14]. In a sample of female participants with ED (N = 25), the participants presented with lower levels of reflective function (RF) regarding the self and higher levels of alexithymia using more emotional suppression and less cognitive reappraisal compared to HCs [12]. Furthermore, in another study by Sacchetti et al. [14], they found that patients with BN (N = 53) differed significantly compared to HC on all mentalizing measures. Specifically, on the Reflective Func­tioning Questionnaire (RFQ) patients with BN scored significantly higher than HC on RFQ-U and significantly lower on RFQ-C (certainty about mental states). Both suggested impaired mentalizing (hypomentalizing and hypermentalizing) in patients with BN, indicating a lack of insight into the mental understanding of self and others. In a study by Gagliardini et al. [15] on adult patients with ED (N = 157), four different profiles of impairments in the dimension of mentalizing were found. The four profiles were heterogeneous in terms of EDs represented in each group and presented significant differences on various variables such as attachment style, emotion dysregulation, empathy, interpersonal reactivity, and reflective function. These findings suggest that patients with ED may be classified in relation to impairments in different dimensions of mentalizing, which suggests a comprehensive assessment should focus on the patients’ mentalizing impairments since patients with the same ED could be characterized by opposite patterns of mentalization impairments. This suggests that clinicians should take a multidimensional approach to mentalization into consideration when treating patients with ED [15]. Further, the suggestion that the MP may be expressed differently in the various ED diagnoses might be important to the future understanding of the MP in ED as well as the implementation of treatment of patients with an ED.

Aim

This study will investigate the MP in patients with an ED by using a cross-sectional survey on a sample of patients diagnosed with ED when compared to HC. Thus, the aim of this study was to investigate whether patients with ED have a lower degree of mentalization ability compared to HC. If the data allows it, a secondary aim would be to investigate whether the MP is expressed differently in AN and BN in an additional exploratory analysis.

The hypothesis of the study is that there is a statistically significant difference in the MP of patients with ED compared with a HC group with patients with an ED having a lower mentalizing function when compared to control subjects. We further hypothesize that patients with AN would have a lower mentalizing function when compared to patients with BN.

Methods

Participants

Participants aged ≥ 18 years were recruited from a specialized hospital eating disorder unit at Psychiatry, Aalborg University Hospital, Denmark. Data was collected over two recruitment periods (October 2017 to February 2018 and April to May 2019). Patients were diagnosed according to the DSM-5 [1] based on assessment with the Eating Disorder Examination (EDE) [16]. HC were recruited through youth education schools and the university in the same geographical area. Inclusion criterion for HC was ≥ 18 years. Known psychiatric diagnosis (self-reported) including ED was considered an exclusion criterion for participation as an HC.

Ethical considerations

Participation for both patients with an ED and the HC was voluntary, and signed informed consent was obtained prior to participation. The patients were informed that nonparticipation would not affect their treatment. The study was approved by the Data Protection Agency, approval number 2017-207. The study was submitted to the Regional Science Ethical Committee which stated that no further approval was needed. It was conducted in accordance with the Helsinki Declaration. To preserve patient privacy in general and, due to the sensitive nature of the study aim, it was decided to follow the regulations set by the Danish Data Protection Agency. Thus, caution was taken for groups consisting of fewer than four patients, and data are not reported in exact numbers but rather shown as informative percentages. To avoid reporting microdata, male participants subsequently had to be excluded due to a very low number. The survey was conducted electronically in REDCap administered by Aalborg University Hospital.

Assessment

Assessment of ED at time of referral patients were diagnosed with a standardized diagnostic assessment battery that included the Eating Disorder Examination, edition 16.0D (EDE) [16,17]. Fully trained clinical psychologists administered the EDE interview after having received formal training before conducting the diagnostic interview. Ongoing supervisions were also provided by a supervisor, PhD level. The EDE assesses the frequency and severity of ED symptoms and behaviors indicative of an eating disorder over a 28-day period. The EDE is scored on a 7-point scale (0–6). Cut-off for meeting a given ED criterion in this study was set at 3. EDE has good internal consistency, discriminant and concurrent validity, and inter-rater reliability [17–19]. Subsequently, at time of inclusion in the study, an additional status of ED pathology was provided by the Eating Disorder Questionnaire (EDE-Q).

An assessment battery was applied in form of a survey, which was identical for both groups. The survey was digitally self-administered in REDCap by all participants.

Assessment of the ED pathology in the survey

The Eating Disorder Questionnaire (EDE-Q) was administered at time of inclusion to provide a status of the ED pathology. Thus, the presence of specific diagnostic criteria and frequency criteria were obtained with EDE-Q, which is a self-report questionnaire with 33 items that assesses symptoms and severity of an ED diagnosis. EDE-Q is developed for clinical diagnosis of ED and has four subscales for assessment of ED symptomatology as well as a global score. The subscales include the restraint, eating, weight and shape concerns. Conservative clinical cutoff for the EDE-Q is most commonly considered to be set at 4 [18,20]. The EDE-Q was tested for overall internal consistency through the Cronbach’s alpha with the data obtained in this study, which gave coefficients between 0.85 and 0.94 for the subscales and 0.96 for the global score. These coefficients correspond to a good or excellent internal consistency.

Assessment of the mentalization profile in the survey

There are several ways to assess for mentalization including test-based assessment, interview-based assessment, and self-report. This study was based on self-reported data with a multifaceted approach corresponding to that applied in the systematic review and meta-analysis by [11]. Several subtests were applied to study the dimensions of the MP. The included subtests were selected as they should (a) meet the requirements of the ability to assess different dimensions of the MP and (b) be appropriate for self-administration (see Table 1). The included test instruments were:

Table 1. Overview of included instruments for assessment dimensions of MP.

Test	Mentalization		Mentalization		Mentalization		Mentalization
	Automatic	Controlled	Internally	Externally	Oneself	Others	Cognitive	Affective
RME		X		X		X	X	X
RFQ-8		X	X	(X)	X	X	X	X
TAS-20		X	X		X		X	X

Overview of the included instruments and their ability to assess the various dimensions of the MP. It should be noted that RFQ-8 only partially assesses the ability to mentalize externally. This table is inspired by an exhaustive list of instruments capable of assessing mentalization by Luyten et al. [10].

Reflective Functioning Questionnaire (RFQ-8), a self-administered test instrument recommended for research on mentalization. RFQ-8 is a 7-point Likert scale questionnaire with 8 items. The questionnaire has two subscales: (A) Items 1–6 (the RFQ-C subscale) assess hypermentalization which is an excessive kind of mentalization that is inaccurate, assumptive and has little or no relationship to reality, and (B) Items 2 and 4–8 (the RFQ-U subscale) assess hypomentalization which describes an inhibited ability to mentalize about mental states in oneself or others. Each item has seven response options ranging from ‘strongly disagree’ to ‘strongly agree’. For the RFQ-C score, response options 1, 2, 3, 4, 5, 6, and 7 are respectively given scores of 3, 2, 1, 0, 0, 0, and 0. For the RFQ-U score, response options 1, 2, 3, 4, 5, 6, and 7 are given scores of 0, 0, 0, 0, 1, 2, and 3, respectively, The RFQ-C and RFQ-U scores are each based on the mean of the 6 items in each score, respectively [8,21–23]. As an official Danish translation of the RFQ-8 was not available, the questionnaire was translated for this study. Both the English and the Swedish version of the questionnaire were independently translated into Danish by two of the authors (CBS and AGJ). The two translations were compared, and choice of words was discussed until agreement was reached in a final version. Subsequently, the Danish RFQ-8 was reviewed by an independent third-party, PhD level. Finally, a back translation was provided as well. The agreement between the back-translated version and the original version was considered good. The RFQ-8, RFQ-C and RFQ-U were tested for internal consistency through the Cronbach’s alpha with the data obtained in this study, which gave coefficients of 0.79, 0.82 and 0.61, respectively. For RFQ-U, the coefficient is rather low, but the reason for this is unknown.

Reading the Mind in the Eyes (RME) test consists of 37 items, each presenting an image of the eye region of a different person. Test subjects are required to choose one of four words that they find best describe the mental state of the person on each image. Only one answer is considered correct, and the global score is computed as the number of correct answers. A low global score on RME indicates low ToM and a low degree of mentalization about others [10,24,25]. By coding the answers to each of the 37 items as correct/incorrect, the Kuder-Richardson Formula 20 (KR-20) of these dichotomous variables was used to compute an overall consistency coefficient of 0.91, which corresponds to excellent internal consistency.

Toronto Alexithymia Scale (TAS-20) consists of 20 items and is an revised version of the original (TAS-26), which consisted of 26 items [26]. The items in TAS-20 are based on a Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). When computing the global score, response options 1–5 are given a corresponding score from 1 to 5 while the scale is reversed on item 4, 5, 10, 18 and 19 and therefore given a score from 5 to 1. The global score is computed as the mean of the scores. A higher global score indicates higher degree of alexithymia and thereby a lower degree of mentalization about oneself [10,26,27]. The TAS-20 was tested for overall internal consistency through the Cronbach’s alpha with the data obtained in this study, which gave a coefficient of 0.89, corresponding to good internal consistency.

Statistical analyses

Descriptive data were produced by use of either means and standard deviations or number of persons and percentages for continuous and dichotomous variables, respectively. The demographic and clinical characteristics were compared using either a two-sample t-test not assuming equal variances or a chi-squared test depending on whether the variable was continuous or dichotomous. Furthermore, the test scores of the HC, patient group and subgroups were compared through mean differences, 95% confidence intervals (CIs) and p-values, which were obtained from linear regression adjusted for the age and with the test score as dependent variable. Whether the p-values are significant on a 5% significance level are generally indicated with stars as follows: p < 0.05: *; p < 0.01: **; p < 0.001: ***. The model assumptions were inspected graphically using Q-Q plots, and linear regression with bootstrapping with 1000 replications (resampling according to the patient groups) were used in sensitivity analyses. The groups’ test scores were furthermore compared through the standardized mean differences (Hedges’ g) not assuming equal variances. Finally, the relationships between the mentalization measures and the eating disorder severity were assessed through partial correlations adjusted for the age.

All statistical analyses were executed using Stata Statistical Software: Release 16. College Station, TX: StataCorp LLC. The figure has been made with Python 3.8.10 through Spyder 5.0.0.

Results

A total of 88 participants, distributed between patients with ED (N = 30) and HC (N = 58), was ultimately included in this study. Twenty-three of the 30 patients met diagnostic criteria for AN or OSFED-AN subtype (N = 23) while the rest met the diagnostic criteria for BN (N = 7) (see Table 2). No significant differences in educational levels and housing situation were found between patients with an ED and the HC. A significant difference was found on age, however, which was partly due to outliers in the patient group. An expected significant difference was found on BMI between the HC and patients with an ED (p = 0.000) for obvious reasons.

Table 2. Demographic and clinical characteristics.

Demographic characteristics
	HC (N = 58)*	ED (N = 30)*	p-Value (HC vs. ED)
Age in years (mean (SD))	21.38 (3.92)	27.30 (8.85)	0.001**
Education level (n (%))
- Primary school, 10th grade, high school	28 (48.28%)	13 (43.33%)	0.315
- Bachelor’s or master’s degree	20 (34.48%)	15 (50.00%)
Housing situation (n (%))
- Living alone	12 (20.69%)	8 (26.67%)	0.802
- Living with others	36 (62.07%)	21 (70.00%)
Clinical characteristics
ED symptomatology	HC (N = 58)*	ED (N = 30)*	p-Value (HC vs. ED)	AN (N = 23)*	BN (N = 7)*	p-Value (AN vs. BN)
BMI (mean (SD))	24.27 (5.93)	19.71 (2.54)	0.000***	19.07 (2.36)	21.82 (1.98)	0.010*
Duration of treatment (n (%))
- Less than 12 months	0 (0.00%)	14 (46.67%)	NA	10 (43.48%)	4 (57.14%)	0.590
− 12 months or more	0 (0.00%)	15 (50.00%)	NA	12 (52.17%)	3 (42.86%)
EDE-Q test scores (mean (SD)):
Restraint subscale	1.46 (1.48)	3.68 (1.91)	0.000***	3.61 (2.16)	3.91 (0.73)	0.567
Eating concern subscale	1.18 (1.34)	3.39 (1.58)	0.000***	3.37 (1.70)	3.46 (1.25)	0.890
Weight concern subscale	2.24 (1.74)	4.28 (1.82)	0.000***	4.03 (1.98)	5.11 (0.72)	0.037*
Shape concern subscale	2.60 (1.78)	4.83 (1.58)	0.000***	4.65 (1.68)	5.39 (1.08)	0.187
Global EDE-Q test score	1.87 (1.45)	4.04 (1.63)	0.000***	3.92 (1.81)	4.47 (0.80)	0.262

Demographic and clinical characteristics on patients with an ED and HC. ED corresponds to all patients. *It should be noted that N varies slightly due to minor occurrences of missing data on educational level, duration of treatment, eating concern subscale, weight concern subscale and shape concern subscale. Specific figures for missing data are not reported due to microdata. Regarding housing situation, there were missing data on ten HCs.

As displayed in Table 2 on demographic and clinical data on ED, symptomatology showed a global EDE-Q score of 4.04 (SD = 1.63) in patients with ED, 4.47 (0.80) in patients with BN, 3.92 (1.81) in patients with AN, while HC had a score below cut-off with a mean global score of 1.87 (1.45) (Table 2).

As can be seen in Table 3, the patients with an ED scored significantly higher on RFQ-U (p = 0.048) and TAS-20 (p = 0.019) compared to the HC. Similarly, the BN group scored significantly higher on RFQ-U (p = 0.023) and TAS-20 (p = 0.006) compared to the AN group.

Table 3. Mentalization profile.

Mentalization profile HC vs. ED
	HC (N = 58)	ED (N = 30)	HC vs. ED (mean difference and 95% CI)	p-Value
RFQ-C	0.97 [0.74; 1.21]	1.09 [0.76; 1.42]	0.12 [−0.30; 0.54]	0.577
RFQ-U	0.70 [0.53; 0.88]	1.02 [0.77; 1.26]	0.31 [0.00; 0.62]	0.048*
TAS-20	2.34 [2.14; 2.54]	2.78 [2.49; 3.07]	0.44 [0.08; 0.81]	0.019*
RME	23.10 [20.86; 25.34]	25.68 [22.46; 28.89]	2.58 [−1.51; 6.67]	0.213
Mentalization profile AN vs. BN
	AN (N = 23)	BN (N = 7)	AN vs. BN (mean difference and 95% CI)	p-value
RFQ-C	1.15 [0.77; 1.53]	1.01 [0.31; 1.70]	−0.14 [−0.94; 0.65]	0.712
RFQ-U	0.83 [0.54; 1.13]	1.55 [1.01; 2.08]	0.71 [0.10; 1.32]	0.023*
TAS-20	2.51 [2.24; 2.79]	3.40 [2.86; 3.95]	0.89 [0.28; 1.50]	0.006**
RME	27.12 [24.58; 29.65]	23.62 [19.02; 28.21]	−3.50 [−8.75; 1.75]	0.183

The mentalization profile of HC vs. ED and AN vs. BN, respectively, with model estimates and CIs as well as contrasts, CIs and p-values, which are all computed using linear regression adjusted for the age. ED corresponds to all patients.

The model assumptions of normally distributed and homoscedastic residuals were seen to be violated to some degree for most of the scores in Table 3, but the results according to the sensitivity analyses were generally similar to those presented in Table 3. However, the p-value for the comparison of HC vs. ED for RFQ-U was 0.058 in the sensitivity analyses, making the mean difference not entirely significantly different from 0.

Figure 1 shows Hedges’ g (i.e. standardized mean differences) and the corresponding CIs. Large effect sizes are seen for RFQ-U and TAS-20 for AN vs. BN, where the Hedges’ g is smaller than −1.

Figure 1. The dimensions of the MP between HC and ED and between AN and BN. Hedges’ g and CIs of the dimensions of the MP between HC and ED and between AN and BN.

Furthermore, it was examined through a linear model adjusted for age whether BMI ≤ 18.5 was related to inhibited MP; however, no significant difference was found when compared to HC.

Finally, the partial correlations between the mentalization measures (RFQ-C, RFQ-U, TAS-20 and RME) and the eating disorder severity (EDE-Q global score) adjusted for the age are shown in Table 4.

Table 4. Partial correlations.

	HC	ED	AN	BN
RFQ-C	−0.15 (0.278)	0.24 (0.205)	0.28 (0.200)	0.35 (0.500)
RFQ-U	0.43 (0.001**)	0.27 (0.164)	0.15 (0.505)	0.80 (0.054)
TAS-20	0.55 (0.000***)	0.23 (0.233)	0.20 (0.381)	−0.12 (0.848)
RME	0.05 (0.686)	−0.05 (0.781)	−0.01 (0.975)	−0.33 (0.564)

Partial correlations between the mentalization measures and the eating disorder severity, measured by the EDE-Q, adjusted for the age, and stratified between the HC, the patient group, and the subgroups (with p-values in parentheses).

As Table 4 shows, there are partial correlations of 0.43 and 0.55 with significant p-values for RFQ-U and TAS-20 compared to the global score of EDE-Q for the HC (p = 0.001 and p = 0.000, respectively).

Discussion

The literature on the MP in patients diagnosed with ED is as sparse as it is ambiguous. This is important as MBT treatment of ED has become increasingly popular in recent years [28] though it is primarily based on a theoretical association between ED and mentalization [4,29–31]. The aim of this study was to investigate the MP in patients with an ED by using a cross-sectional study design on a sample of patients diagnosed with ED compared to HC. We aimed to investigate whether patients with ED have a lower degree of mentalization ability when compared to HC, and, if possible, to investigate whether the MP is expressed differently in patients with AN and BN. The study results showed statistically significant differences in RFQ-U and TAS-20 when comparing patients with ED to HC, which corresponds to the results of previous reviews and meta-analyzes that point towards a reduced mentalization ability about the self (assessed with TAS-20) in individuals with ED when compared to HC [11,32,33].

As described in Background, the ability to mentalize is multifunctional, and concepts such as ToM and alexithymia are both capable of describing components of the MP. As described, ToM refers to the ability of understanding mental states of others and thereby describes the dimension of the MP that concerns mentalization about others. ToM can be assessed using the RME included in this study. To the best of our knowledge, so far only one study has produced a systematic review and metanalysis on the MP in patients diagnosed with ED [11]. While, Simonsen et al. [11] found mixed results when reviewing studies of ToM in ED when compared to HC assessed by the RME, the study concluded that there was no general difference in the ability to mentalize about others when comparing ED and HC [11]. Therefore, the lack of significant difference when comparing patients with ED with HC in this study seems to be in accordance with earlier findings indicating that patients diagnosed with ED do not differ from HC in ability to mentalize about others.

Another psychological concept highlighted in this study is alexithymia which refers to the inability to understand, identify and describe own emotional states. Alexithymia can be assessed by TAS-20, and in this study, as mentioned, the level of alexithymia was found to be significantly higher in the BN patient group when compared to AN. As mentioned above, studies of alexithymia in patients with ED have indicated the ability to mentalize about the self to be less impaired in patients diagnosed with BN compared to patients with AN. Unexpectedly, the results of this study do not repeat the previous findings by Simonsen et al. [11] who found particularly inhibited mentalization about the self in AN, as the study findings point towards impaired mentalization about the self in the BN group. However, it should be emphasized that these findings should be taken with the utmost caution due to the small sample size of BN patients.

While ToM was assessed with RME and alexithymia was assessed with TAS-20 in this study, RFQ-8 does not represent a specific dimension of the MP. RFQ-8 was included in this study because of its capability to describe mentalization broadly across different dimensions. A significant difference was found between AN and BN when assessing mentalization with the RFQ-U subscale. This finding indicates a lower ability to mentalize across different dimensions of the MP in general as well as a higher degree of hypermentalization in patients with BN compared to AN. However, as previously mentioned, the present findings of a statistically significant difference on both TAS-20 and RFQ-U between AN and BN should be treated with great caution due to the very small sample size of patients with BN included in this study.

As described, there is a wide range of testing tools that can assess specific but not all dimensions of the mentalization ability. The study included widely used and reliable instruments to describe the contextual and multidimensional MP. However, none of the test instruments included could assess automatic mentalization as self-administered tests require conscious consideration when answering. That is why all included tests assess controlled mentalization [10]. Future studies of automatic mentalization in patients with ED could be relevant. The mixed results of both this study and earlier studies on specific dimensions of the MP in individuals diagnosed with different EDs may be explained by the fact that the MP is particularly contextual. In this study, it was decided to include a self-administered test battery containing test items that were designed to investigate mentalization in a more general context as opposed to the Reflective Functioning Scale which is typically applied to an Adult Attachment Interview that focuses on attachment contexts [10,34]. It is reasonable to assume that while the mentalization ability is quite well-functioning in individuals with an ED while sitting at home in front of the computer answering questionnaires, as is the case with this study as well as most others, the ability to mentalize in contexts that include eating may be far more inhibited in individuals suffering from an ED. Therefore, further research on the ability to mentalize in eating contexts in individuals diagnosed with ED is needed.

Strength and limitations

This study is one of only a few with a multidimensional approach to assess the mentalization profile in patients with ED. However, some limitations should be mentioned. First and foremost, it should be emphasized that the study has a relatively small sample size. Therefore, advantageously more participants in the patient group could have been included, not least regarding patients with BN. Also, no power calculations were made before the data collection. The self-administered test battery made it impossible to control the contexts in which the reporting was conducted. Unfortunately, it was deemed necessary to exclude male participants due to microdata. The participants were furthermore not assessed for potential co-morbidities or use of medication. A further limitation that should be mentioned is that the groups were not matched on age. Finally, some concerns related to the psychometric properties of the RFQ8 should be mentioned as a potential limitation as studies suggest that the RFQ8 has a unidimensional factor structure [35,36]. Future research should focus on MP and include a larger sample size as well as male participants.

Conclusion

The aim of this cross-sectional study was to investigate the association between the MP and ED in a sample of patients diagnosed with ED when compared to HC. The purpose was to investigate whether patients with ED have a lower degree of mentalization ability when compared to HC and whether the MP is expressed differently in patients with AN and BN.

These study findings indicate that patients with an ED do have some impaired abilities to mentalize compared to HC. Thus, the study results show a statistically significant difference between the patients with ED and HC with patients with an ED scoring significantly higher on the RFQ-U and TAS-20 than the HC. Although the results indicate that patients with BN may have a lower ability to mentalize across different dimensions of the mentalization profile when compared to patients with AN, the results of this study should be interpreted with great caution due to the small sample size in the BN group.

Further research should be conducted to gain knowledge on potential clinical support with respect to the theoretical basis for mentalization based therapy in the treatment of eating disorders.

Acknowledgments

We would like to thank the participating patients and controls for participating in this study. Thank you to the staff at the Unit for Eating Disorders, Aalborg University Hospital for supporting the study. Finally, thank you to Birgitte Christiansen and Christina Kjaer Frederiksen from the Unit for Psychiatric Research, Psychiatry, Aalborg University Hospital for invaluable assistance with this publication.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Due to the ethical, privacy and security conditions applicable to this manuscript, data is not shared. Further information regarding data can be accessed by contacting the corresponding author.