Introduction
Insight refers to the individuals’ understanding and awareness of their illness, including the severity of their symptoms and the impact of their behavior on their health and well-being 1,2. The concept of insight has acquired an ever greater significance in the context of modern psychopathology 3,4. Several conditions with an organic or functional basis are characterized by an unusual lack of insight, which manifests itself either as a total absence of awareness of the disease (referred to in the neurological literature as “anosognosia”) 5,6, or as a deficit of insight, in particular as regards the impact of the disease on the state of well-being and psychosocial adaptation 7. The consequence is often the partial or total refusal of acceptance of the role of patient, and therefore of adherence to therapeutic protocols, or the expression of anomalous behaviors in assuming the role of patient 7,8. In psychiatry, insight deficit up to anosognosia is considered a hallmark of schizophrenia (p. 117) 9, while different degrees of insight deficit are found in obsessive-compulsive disorder, currently defining the level of severity of the disorder and associated syndromes (F42.2; F45.22; F42.3) 9.
People with anorexia nervosa, too, may have varying levels of insight into their illness. Some may recognize that they have a problem and seek treatment willingly, while others may deny or minimize the severity of their symptoms, resist treatment, or be unaware of the consequences of their behaviors 10-12. This defective awareness of illness during an eating disorder, although known for decades 13,14, has not been specifically addressed except in recent years 15, related to the expanding interest in the study of insight in psychiatry 4. As a result of this poor thematization of the concept of insight in eating disorders, its measurement has mostly been based on clinical impressions or on the use of self-compiled tools developed for other purposes. This has led to a wide variation in estimates of the phenomenon in samples including patients diagnosed with anorexia nervosa, from 15% to 80%, suggesting that these methods are poorly reliable 15.
A tool primarily aimed at measuring the degree of insight in eating disorders has recently been developed, the Schedule for the Assessment of Insight in Eating Disorders - SAI-ED. The SAI-ED is a semi-structured interview that measures three major components of insight, such as the ability to recognize that one has a mental illness, the ability to relabel unusual mental events as pathological, and compliance with treatment 15. The SAI-ED has been developed from a pre-existing tool aimed at measuring insight in patients with a mental disorder 16, and has undergone several adaptations over time 15,17. Since it is a clinician-rated instrument and specifically investigates symptoms of eating disorders (e.g., body image distortions or eating behaviors), the SAI-ED is expected to be more accurate in assessing the dimensions of insight in patients with an eating disorder. Moreover, a semi-structured interview reduces the risk of misunderstandings the questions or biases such as the ‘over-compliance’ effect (patients’ eagerness to please resulting in ‘pseudo-agreement’), which is more frequent with self-report questionnaires. Moreover, the study of specific symptoms of eating disorders is crucial given their importance in treatment 18,19.
This is a pilot study aimed at a) verifying the feasibility of the use of the SAI-ED in a clinical sample of patients diagnosed with anorexia nervosa; b) testing the reliability of the tool; c) assessing its convergent validity, by comparison with a self-report measure of insight; d) analyzing the clinical correlates of the tool.
Methods
The study has been conducted according to the Declaration of Helsinki and its revisions and amendments 20,21. This study was approved by the Ethical Committee of the local academic hospital (Prot. n. 0014269; code: 446/2021; signed on February 8th, 2022). According to the approved protocol, all participants signed a written informed consent.
Participants
All consecutive admissions among those seeking voluntary admission for anorexia nervosa to a major academic Eating Disorders Center between March and December 2022 were invited to take part in the study.
Inclusion criteria were: (a) confirmed diagnosis of anorexia nervosa according to the Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders, 5th edition (SCID-DSM-5) 22; (b) age between 18 to 50 years old; (c) no current or past psychotic or bipolar disorder or current substance use disorder; (d) capacity to read Italian language. Among 34 patients who were proposed to take part in the study, 5 refused or were too ill to be involved in the study, 1 was discarded because had difficulties in reading the Italian language, and 4 were older than the age cutoff (> 50 years old). Overall, 24 patients with anorexia nervosa were included in this pilot study: 15 with anorexia nervosa restricting type (AN-R), and 9 with anorexia nervosa binge-eating/purging type (AN-BP).
Measures
The study applied the semi-structured SAI-ED interview to the participants (see Supplementary material). Standard translation and back-translation procedures were used to translate the Italian version of the SAI-ED from the English version 23. The back translation of the SAI-ED has been revised by an English-speaking editor and further checked by an English-speaking psychiatrist who is an expert on eating disorders but was not involved in the translation and the adaptation of the tool. The original authors were not involved in the adaptation of the tool.
In its current version, the SAI-ED includes two questions on awareness of the illness and the need for treatment, rated on a 0-2 scale; four questions on the recognition of symptoms of eating disorder and their relabeling as attributable to the illness, plus a question on the hypothetical contradiction between own’s view of the condition and concerns of other people, all rated on a 0-4 scale. There is also a final question on adherence to treatment, aimed at collecting therapist impressions, rated on a 0-5 scale (see supplementary material). Each section of the schedule has its threshold to decide whether insight was fair to good or partial or absent (see results). The SAI-ED produces two global scores: a subtotal on awareness of illness and symptoms, the sum of the first seven items (ranging from 0 to 24); and a total score, the sum of all items, including the item on the adherence to the treatment (ranging from 0 to 29).
The internal consistency of the SAI-ED was reported to be high, as it was the inter-rater and test-retest reliability 15. Concurrent validity, which was tested with the Brown Assessment of Beliefs Scale (BABS) 24, was also good. In the initial validation study, the lack of insight was likely for a global score on the SAI-ED ≤ 18.
Additional tools used in this study were the Beck Cognitive Insight Scale (BCIS), the Eating Disorder Examination Questionnaire (EDE-Q), the State-Trait Anxiety Inventory (STAI); and the Beck Depression Inventory (BDI).
The BCIS is a 15-item self-report scale, of which 9 items define the self-reflectiveness subscale and 6 items define the self-certainty subscale 25. On each item, participants rate the extent they agree with statements on a scale from 0 (do not agree at all) to 3 (agree completely). A global score of cognitive insight is derived from the two subscales, the Composite Reflectiveness-Certainty Index, by subtracting the score of the self-certainty subscale from the score of the self-reflectiveness subscale. The higher the self-reflectiveness score and lower the self-certainty score, the higher the cognitive insight. The reliability of the BCIS is suboptimal, with the internal consistency of the two subscales ranging between 0.60 and 0.70 26, still within the interval of acceptability for the reliability of a subscale 27. Overall, the two-factor structure of the BCIS has been confirmed in both Western and non-Western samples 28, and evidence of convergent validity has been found with 5 different measures of clinical insight 26.
The EDE-Q is a 28-item self-report questionnaire investigating four main symptom areas of eating disorders: dietary restraint, eating concerns, weight concerns, and shape concerns 29. On each item, the participants rate the intensity of their behavior or thoughts during the last 28 days on a scale from 0 to 6. A global score also is produced. The reliability of the tool is usually good (> 0.80) in clinical samples 30.
The STAI is a 40-items self-report questionnaire that assesses levels of anxiety 31. On each item, the participants rate their levels of anxiety as experienced at that moment (state anxiety, 20 items) or habitually (trait anxiety, 20 items) on a scale from 1 (never) to 4 (always). The reliability of the STAI ranges from 0.65 to 0.90 depending on the sample 32.
The BDI is a 21-item self-report questionnaire that measures the presence and severity of depressive symptoms 33. Participants were asked to rate their behavior or thoughts on a scale from 0 to 3. As higher the scores, as severe the symptoms of depression. Scores above 16 are an index of severe depressive symptomatology. The tool has a high internal consistency, around 0.90 across samples 34.
Procedure
Each participant was invited to sit in a quiet room where they were interviewed with the SAI-ED. The interview had a duration, on average, of 15 minutes. Additional questionnaires were implemented with the REDCap platform (Research Electronic Data Capture, Vanderbilt University), and each participant filled in the questionnaires on a tablet or smartphone. The time to completion of these questionnaires was not recorded. Before starting the assessment, participants were asked for informed written consent, and anonymity was granted.
Data analysis
All data were coded and analyzed using the Statistical Package for the Social Sciences (SPSS) version 28 (IBM Corp. Released 2021). Further analyses were performed in R, using dedicated packages 35.
The significance threshold was set at p < 0.05. All analyses were two-tailed.
Continuous variables were reported as means with standard deviations. Categorical variables were reported as counts and percentages. Analysis of continuous variables was performed with Student’s or Welch’s t-test in case of violation of the homogeneity of the variance. Categorical variables were analyzed with Chi-square, with Yates correction when necessary. Pearson correlation coefficient (aka Pearson’s r) was used to measure the association between continuous variables. Effect size in between-group comparisons and correlations was assessed with reference to the thresholds suggested by Cohen 36.
The reliability of the self-report questionnaires was measured with Cronbach’s alpha to facilitate comparisons with previous studies. According to a shared rule, Cronbach’s alpha is considered “fair” to “acceptable” when it exceeds values of 0.60 in subscales of a multidimensional questionnaire; it is assumed to be “good” when it is equal to or greater than 0.70 in a questionnaire measuring one latent dimension 27.
Reliability of the SAI-ED was assessed as agreement between independent raters using Cohen’s kappa for individual items of the SAID-ED and as intraclass correlation coefficient (ICC) with 95% confidence interval (CI) for the global scale of the SAID-ED. Cohen 37 suggested interpreting values 0.20-0.40 of kappa as indicating “fair”, 0.41-0.60 “moderate”, and 0.61-0.80 “substantial” agreement. According to a shared rule of thumb, ICC values between 0.60 and 0.75 indicate moderate reliability, values between 0.75 and 0.90 indicate good reliability, and values above 0.90 indicate excellent reliability 38.
Results
The patients were 23 women and 1 man and had a mean age of 27 years (standard deviation [SD] = 9; age range from 19 to 49 years). Body mass index (BMI) was 14.5 in the sample (SD = 1.8), with AN-R having a lower BMI than AN-BP: 13.9 ± 1.6 versus 15.7 ± 1.6 (t = 2.39; df = 22; p = 0.013).
Reliability of the measures used in the study
For a subset of 10 subjects, the SAI-ED was assessed by two independent raters. Cohen’s kappa values were 0.704 (p = 0.024) for the first item; -0.143 (p = 0.686) for the second item; 0.797 (p = 0.005) for the third item; 0.871 (p = 0.008) for the third item bis; 0.544 (p = 0.010) for the fourth item; 0.783 (p = 0.002) for the fourth item bis; 0.579 (p = 0.009) for the fifth item. With the exception of the second item, assessing awareness for the need of treatment, there was substantial agreement between the raters. The ICC of SAID-ED in the total sample of participants diagnosed with anorexia nervosa was 0.711 (95%CI: 0.490-0.860). Overall, the interview had moderate reliability.
The BCIS self-reflectiveness scale had a Cronbach’s alpha of 0.818, while the Cronbach’s alpha of the BCIS self-certainty subscale was 0.748. Cronbach’s alpha for EDE-Q was 0.890; it was 0.951 for the STAI, and 0.969 for the BDI. Overall, reliability of self-report questionnaires in the sample was good.
Distribution of the scores of the anorexia nervosa participants on items of the SAI-ED
Figure 1 summarizes the distribution of the scores of the participants on each item of the SAI-ED. The first two items on the recognition of having a mental illness and the need for treatment have a scale of 0 to 2, with 2 being the threshold for presence of awareness. As it can be seen, participants were more ready to accept they need treatment than to recognize they have a mental disorder. The other items have a scale of 0 to 4, with 3 being the threshold for presence of moderate awareness, and participants were more willing to recognize body size concerns than eating pathological behavior as attributable to their illness. Overall, their propensity to recognize the hypothetical contradiction in their beliefs with respect to others’ concerns was limited, with scores well below the minimum score for moderate awareness.
The treatment engagement scorer anged from 0 to 5, and the participants’ mean score was 4.2 ± 0.9, equivalent to a moderate acceptance of the prescribed treatment.
The participants’ score on the subtotal on awareness of illness and symptoms was 17.6 ± 4.1 (range: 9 to 24). The participants’ overall mean score on the SAI-ED was 21.9 ± 4.4 (range: 12 to 29). In the sample, less than one third (27%) scored≤18 on the SAI-ED, the threshold for poor insight. Overall, insight in this pilot sample was fair to moderate and never full. The scores on the awareness of illness and symptoms subscale were statistically associated to the therapists’ rating of adherence to treatment: r = 0.494, p = 0.014.
Association of the scores on the SAI-ED with clinical characteristics
Global scores in the SAI-ED were positively associated with levels of depression on the BDI (Tab. I).
No other links emerged.
Patients diagnosed with AN-R reported lower insight on the SAI-ED compared to patients diagnosed with AN-BP: 18.8 ± 5.3 versus 22.3 ± 1.7; Welch’s t = -2.1; df = 12.3; p = 0.036. There was also a positive relationship with BMI: r = 0.523, p = 0.0008, i.e., increasing BMI was associated with increasing insight.
Discussion
The Italian version of the SAI-ED in this sample showed a moderate reliability, with lower values than those reported in the initial validation study. The difference might depend on the sample size, which was smaller in our study (n = 24) than the original validation study (n = 44) 15.
Scores on the SAI-ED were significantly lower in patients with AN-R, hence they had lower insight into illness, than in patients with AN-BP. Patients with AN-R are more likely to perceive their symptoms and behaviors as ego-syntonic than patients who binge and purge 39. Patients who binge and purge are afraid of becoming fat, while purely restricting patients are moved by a drive for thinness that is maintained by restrictive eating. The capacity of the SAI-ED to distinguish patients with different levels of insight is congruent with past studies that compared the level of insight between anorexia nervosa and bulimia nervosa using self-report SAI-ED 17, BABS 40, and the interview-based SAI-ED 15.
In this sample, greater impairment of insight was associated with a lower current BMI. Previous studies also found in anorexia nervosa significant associations of insight with current BMI in the same direction we have found 15,41. Patients with AN-R had lower BMI than patients with AN-BP, and this might partially explain the relationship between poorer insight and lower BMI. However, it cannot be ruled out the effect of starvation and consequent malnutrition on cognition. One long-term follow-up study on insight in anorexia nervosa found that an improvement in BMI was related to higher insight at follow-up 42.
We did not find a link between the levels of insight and severity of symptoms on the EDE-Q. In the previous application of the SAI-ED to a sample of 44 patients with anorexia nervosa, Konstantakopoulos et al. 15 found a negative association between global scores on the SAI-ED and the restraint subscale of the EDE-Q. Sample size probably is the main reason for the discrepancy: as said, we had a smaller sample than in the Konstantakopoulos et al. study 15.
Contrary to the results of previous studies on eating disorders 14,15,17,41, we found an association between insight and depression. Depressive symptoms are common in EDs and may have different origins 43,44. In past studies, depression was measured as categorical 14, or with the Hospital Anxiety and Depression Scale (HADS) 45. We used the BDI, and the different methods to measure depression might contribute to explaining the discrepancy between ours and past results.
An association between insight and depression was found in both Alzheimer’s disease and schizophrenia 46,47. It might be surmised that awareness of the severity of their condition and its consequences is likely to cause a depressive reaction in people with schizophrenia or early stages of Alzheimer’s disease. Conversely, at least in Alzheimer’s disease, impaired awareness of deficits seems to be related to fewer depressive and anxiety symptoms 48. Nevertheless, it cannot be excluded that the occurrence of depression, resulting in more realistic self-evaluation of themselves and one’s mental health, might favor the concurrent onset of insight 49. In patients with anorexia nervosa, who value positively their condition 39, the onset of awareness of the pathological nature of their behavior and beliefs might likely induce depression if accepting treatment implies losing the emotional and psychosocial benefits deriving from their condition. Since it is known that depressive symptoms and distortion of the body image are closely connected 43, we can surmise that the severity of depressive symptoms may be correlated with the severity of denial of one’s emaciation, thus with poor insight.
It has already been suggested that a fraction of patients with anorexia nervosa, albeit aware of the pathological nature of their disorder, might be reluctant to express this awareness to maintain control over their eating behavior or to avoid stigma 17. This propensity to active denial of symptoms of both eating disorders and anxiety and depression, observed in some past studies 50, might partially explain our findings of a link between awareness of illness and depression. Patients who were willing to express awareness of illness also would report symptoms of depression.
The lack of association of the scores on the SAI-ED with those on the BCIS might depend on the BCIS having been developed to assess the deficit of insight in psychosis. This would reinforce the need for a measure focused on the disorder-specific symptoms to measure insight in people with eating disorders. We didn’t find a past application of the BCIS to people with anorexia nervosa and it has been questioned whether the BCIS might be valid and useful for patients with non-psychotic disorders 3. Nevertheless, the association of the scores on the awareness of illness and symptoms with the therapists’ rating of treatment adherence provides evidence of the internal validity of the SAI-ED.
Some limitations of this pilot study should be considered. We had a small sample size, which prevented a more articulate analysis of the findings (for example, about the role of comorbidity). The study was planned as a pilot study to verify the feasibility of the use of the SAI-ED in a clinical sample of patients diagnosed with anorexia nervosa, who may be challenging to evaluate on a sensitive topic such as their awareness of illness. Nevertheless, a sample size of 25 participants is a major limitation of the study.
Additional limitations should be considered. We only had patients with anorexia nervosa, and in future studies, a comparison with patients with bulimia nervosa is advisable to confirm that deficit of insight is a feature of anorexia nervosa. Except for one male patient, all participants were women. Hence, the generalizability of the findings to male patients with AN cannot be guaranteed. The sample was recruited in the hospital section of our center, thus including patients severe enough to require hospitalization but also likely to have some insight into the need for treatment. However, mean scores on the SAI-ED in our study were comparable to those found in the original study 15.
Conclusions
The use of a dedicated semi-structured interview is an advantage in measuring insight in patients with an eating disorder. Impairment of insight contributes to the avoidance of treatment or refusal to adhere to treatment and might increase the chance of dropping out and consequent relapse. There is some evidence that impaired insight predicts poor outcomes in patients with anorexia nervosa 51. Thus, improvement of insight might represent a marker of response to treatment. Within this perspective, the Italian version of the SAI-ED is a reasonably reliable, valid, and usable tool for the multidimensional assessment of three critical components of insight and it is devoid of typical weaknesses of previously used instruments.
Acknowledgements
The authors want to dedicate the study to the late Dr. Enrica Marzola, who contributed to ideating and planning of the study. The Eating Disorders Unit staff of the University of Turin heartily remembers Dr. Marzola.
Conflict of interest statement
The Authors declare no conflict of interest
Funding
This paper received no external funding.
Author contributions
AS, AP, and GAD contributed to the conceptualization of the study. AS, MM, and AP organized the study. AS, FB, AP, ML, FT, LA, MP, MM, and AM contributed to participant recruitment and data collection. AP, AS, FB, MM, and GAD analyzed the data. AS, FB and AP wrote the original draft of the manuscript. MP, MM, and GAD provided critical revision of the manuscript.
Finally, all the authors have approved the final version of the manuscript
Ethical consideration
All procedures performed in studies involving human participants were in accordance with the 1964 Helsinki declaration and its later amendments. This study was approved by the Institutional Ethics Committee (Prot. n. 0014269; code: 446/2021; signed on February 8th, 2022). Informed consent was obtained from all individual participants included in the study.
History
Received: April 30, 2023
Accepted: January 25, 2024
Figures and tables
FIGURE. 1. Scores of the anorexia nervosa participants on items of the Italian Schedule for the Assessment of Insight in Eating Disorders (SAI-ED). Dashed gray lines indicate the threshold for good and acceptable insight. Mean scores by item are reported above the corresponding barplot.
| Item | SAI-ED |
|---|---|
| Beck Cognitive Insight Scale (BCIS) | |
| Self-reflectiveness | r = 0.154; p = 0.472 |
| Self-certainty | r = 0.394; p = 0.057 |
| Cognitive insight composite index | r = -0.095; p = 0.659 |
| Eating Disorder Examination Questionnaire (EDE-Q) | |
| Restraint | r = 0.247; p = 0.244 |
| Eating concern | r = 0.256; p = 0.227 |
| Shape concern | r = 0.276; p = 0.192 |
| Weight concern | r = 0.165; p = 0.441 |
| State-Trait Anxiety Inventory (STAI) | |
| State anxiety | r = 0.105; p = 0.625 |
| Trait anxiety | r = 0.140; p = 0.514 |
| Beck Depression Inventory-II | r = 0.426; p = 0.038 |
PDF
