A factor analysis of the meanings of anorexia nervosa: intrapsychic, relational, and avoidant dimensions and their clinical correlates

Notwithstanding the increased knowledge achieved in recent years on eating disorders, anorexia nervosa (AN) still represents a difficult to treat disorder. In fact, AN sufferers often refuse treatments [1], show poor compliance with therapy leading to high dropout rates [2], relapse [3], and high mortality [4]. A deeper understanding of those factors underpinning patients’ difficulties with treatments is thus relevant to improve clinicians’ therapeutic approach to this challenging disorder.

Consensus has been reached on the adaptive function of eating symptomatology; in fact, starvation would help patients avoid negative emotionality [5, 6], reinforce their identity [7, 8], and express their distress [9]. Consistent with these lines of research, AN symptoms would assume a “pro-AN” function in turn maintaining the disorder [10].

The aforementioned hypotheses are in line with cognitive-behavioral models of AN maintenance [10‐12] as well as with psychodynamic models that assess defense mechanisms, ego-syntonicity, and compensation for eating symptoms [13‐17]. Such theoretical models [10, 14] rely on a handful of studies investigating patients’ meaning of AN [7, 12, 18, 19]; moreover, the majority of these papers used either qualitative, descriptive, or phenomenological methods. Using interviews or focus groups these studies identified some meanings that are likely to be attributed to AN by sufferers. For example, Nordbø and coworkers [7] identified eight main constructs for AN encompassing security, avoidance, mental strength, self-confidence, identity, care, communication, and death. Williams & Reid [12] highlighted that ambivalence towards treatment correlates with patients’ description of AN not only as a disease but also as a tool and a way to achieve their own identity. Such ambivalent features have been sometimes described as the “anorexic voice” speaking to patients’ recovery oriented parts [12, 18]. Our group has previously identified difference, company, and identity as adaptive areas of AN; notwithstanding, ambivalence towards the illness and negative sequelae of AN were also reported by sufferers [19]. A review of 24 qualitative studies [8] confirmed the existing findings suggesting that patients highly value their preoccupations with food and weight. Moreover, this paper identified some factors that are linked to the meaning of AN and then grouped them into two meta-categories: need of control and identity [8].

To date, only a dearth of quantitative studies measured pros and cons of AN from a subjective standpoint, for example analyzing letters written by individuals with AN to their own eating disorder [20]. Relatedly, the Pros and Cons Anorexia Scale (P-CAN [21, 22]) and then the Pros and Cons Eating Disorders Scale (P-CED [23]) have been proposed. These instruments confirmed that AN sufferers experience a variety of positive feelings towards their illness like safety, identity, and being special, only to name a few. These elements could underpin patients’ ambivalence towards recovery and become less valuable when patients start to improve their clinical condition [24]. However, several aspects of AN have been so far not in-depth investigated, like the role of avoidance in AN and the relationship between the reported meanings of AN and patients’ clinical characteristics like personality, duration of treatment and, most importantly, duration of illness. Still, we attempted to develop a brief instrument since it can be of help in clinical practice. Therefore, the rationale for conducting this study is grounded on these elements.

Therefore, the overarching aim of this study was two-fold: a) to perform a factor analysis of a new measure testing its reliability and validity; b) to verify the correlations of this measure with eating disorder severity, personality, and duration of illness or duration of treatment.

We hypothesized that the new measure would have reliably captured AN meanings in an easy-to-administer way and that certain subjective meanings of AN could be a constitutive element of illness in turn involved in maintaining the disorder. Therefore, we expected to find a correlation between certain meanings of illness and clinical data (particularly personality) independently of duration of both illness and treatment.

This study was conceived to present and use a novel instrument (MANQ) aimed to quantitatively analyze the potential triggers for AN, the meanings AN sufferers attribute to their illness, and the negative consequences of AN. With respect to the meanings of AN, our research identified three different factors which are related to patients’ personality and motivation to treatments but unrelated to duration of both illness and treatments. Taken together, our findings also confirmed previous literature, mostly with respect to triggers and AN-related negative effects [7, 8, 19].

With respect to the statistical analysis, the scale of the MANQ on the meanings of AN reported an overall solid statistic validity (Cronbach’s alpha = 0.853). Moreover, the section on meanings showed good validity indexes (i.e., capturing 65 % of the variance) when investigating the independent factors (i.e., intrapsychic, relational, and avoidance factors) underpinning the 12 questions of this section. Similarly, also the three identified factors resulted to be psychometrically robust (Cronbach’s alpha = 0.853, 0.705, and 0.795, for factors 1, 2, and 3, respectively). Therefore, this novel instrument may have clinical implications since it is not only statistically sound but also straightforward and easy to administer: in fact, only 5 min are required to complete all questions.

It is of note that the vast majority of patients could identify those elements that elicited the onset of their illness thus trying to offer an explanation for this moment of their lives. According to previous literature, patients were likely to report body dissatisfaction, teasing about weight and shape [33], and problems of adolescence [34] as potential triggers for the onset of AN. Still, those with a binge-purging subtype of AN reported significantly more often than those affected by a restricting subtype of AN sexual traumas, as expected given the existing literature on this topic [35, 36].

Similarly, we found data in line with previous literature also with respect to the subjective meanings of AN [8, 19]. For example, a broad consensus emerged about AN as a way to be protected or to communicate. It is noteworthy that AN-BP patients expressed significantly more frequently than those with AN-R that their eating disorder is a way to give up living; this is consistent on one hand with the data of the present paper on depressive symptoms reported by the two clinical subgroups and on the other hand with the existing body of evidence. In fact, those affected by the AN-BP subtype are known to be often plagued by severe psychopathology [37‐39] and depression [40] and to describe their disorder as a source of safety more often than the AN-R group, potentially mirroring a worse prognosis [3, 41]. Longitudinal studies are worth assessing whether patients tend to score differently on this item in case of diagnostic cross-over or positive outcome. Also, it should be acknowledged that previous lines of research (i.e., P-CAN [21, 22], P-CED [23]) focused on the meanings of AN. Nevertheless, the MANQ could be used in the light of feasibility, given its easiness of administration (i.e., only 12 questions); still, this instrument pays close attention to the avoidance element providing also a broad range of possibilities of response (e.g., Likert-format responses). Moreover, this questionnaire captures also two different sections: the first one on triggering factors and the second one on the consequences of illness. Therefore, depending on the research questions, different instruments could be used in order to shed light on the multifaceted aspects of AN.

Being controlled by the disorder is most frequently reported as a negative consequence of AN, as already reported by our group [19]. Such an element can be useful in treatment: in fact, when patients realize to be overwhelmed and trapped by the disorder they somehow start to criticize it alike. Therefore, this may represent a good starting point in therapy and further research on this topic is recommended [42, 43].

Still, the factorial analysis conducted on patients’ meanings of AN provided interesting data. Three factors, explaining 65 % of the variance emerged and were labelled as intrapsychic, relational, and avoidance factors. Relatedly, a qualitative study conducted with recovered individuals reported how relevant can be overcoming the idea of AN as a way to perform control highlighting instead the need for improving relational skills and emotional management [44]. It is of interest that in our study these factors emerged as clearly separated: only the item assessing need for recognition was shared by both intrapsychic and relational factors, possibly because it could be referred to both personal and others’ recognition. Such an independence of the three factors could have clinical implications in the light of improving highly individualized treatments [10]. Accordingly, we found an inverse correlation between the relational factor and age of onset: in fact, it is well-known how both family- and communication-related elements can be relevant in treatment, particularly with adolescents [45]. Illness denial correlated with the intrapsychic and relational factors in a direct and inverse way, respectively. This finding is of interest since the intrapsychic factor seemed to be more related to the ego-syntonic aspects of AN while the relational factor would indicate greater awareness of illness, including the communication adaptive function of the disorder. Consistently, the intrapsychic factor inversely correlated with motivation to treatment; future studies are needed to investigate as to whether the relational factor (i.e., greater awareness of illness) could correlate with a favorable prognosis, even more so because ego-syntonicity and denial of illness are intertwined with resistance to treatments [1].

Avoidance was the most focused factor; however, it resulted to be independent of the intrapsychic one. A number of models described avoidance as an independent mechanism which is key, at least in a subgroup of patients, with respect to both onset and maintenance of AN [5, 6, 46‐48]. The positive correlation between this factor and patients’ age and number of previous hospitalizations would require further investigations to ascertain whether avoidance can play a role in adulthood and in treatment planning.

All factors correlated with eating psychopathology as measured by the EDI-2 with no specific patterns of correlations. However, the greater the EDI-2 scores the greater the scores of the factors; this finding provides further support to the “adaptive function” of AN in the maintenance of the disorder [10]. Also, the more severe was the eating disorder the more patients seemed to value safety and structure somehow provided by AN [23]. Instead, when running multivariate regression analyses, the vast majority of correlations were not confirmed; however, the intrapsychic factor resulted associated with a core aspect of eating psychopathology, namely drive for thinness and the relational factor with impulse regulation.

With respect to personality, all factors inversely correlated with SD. Since low SD is suggestive of a fragile identity and difficulties with pursuing goals [30] this kind of correlation would raise the possibility that patients may need the disorder to help them shape their identity, as previously suggested [7, 8, 19]. Moreover, both intrapsychic and avoidance factors correlate with HA; this provides support to our a priori hypothesis, given the relevance of HA for AN patients [49‐51] with multivariate regression analysis overall confirming this trend of association.

Finally, the last aim of this study was to determine the independence of subjective AN meanings from duration of both treatment and illness. The positive meanings of AN resulted to be independent of all other elements since no correlations were found with either duration of illness or treatment/psychotherapy. Therefore, the scarring effect of the illness could be excluded as well as an “over-rationalizing” effect of psychotherapy [9]. From a clinical standpoint this is encouraging since patients sometimes perceive to be “taught” what to feel and think in treatment, regrettably confirming their feelings of ineffectiveness and hindering a real awareness of their psychological resources [9].

This study suffers from some limitations: the questionnaire is not validated, the sample size of the AN-BP subgroup is modest, and to quantitatively measure patients’ opinions can be theoretically questioned. Nevertheless, our data are overall in line with qualitative studies on this topic and highlight some novel elements like the relationships between meanings of AN and illness severity and personality, independently of the treatment received.

In closing, this study highlights that patients with AN frequently attribute a subjective meaning to their disorder and even more often tend to feel trapped and overwhelmed by the disorder itself. Such meanings correlate with both illness severity and patients’ character, independently of duration of illness and treatment. Sufferers’ perspectives and opinions on their disorder should be carefully considered in treatment because of their usefulness in both overcoming resistance [43] and engaging patients in treatment [52].

AN, anorexia nervosa; AN-BP, anorexia nervosa binge-purging subtype; AN-R, anorexia nervosa restricting subtype; ANSOCQ, Anorexia Nervosa Stages of Change Questionnaire; BDI, Beck Depression Inventory; BMI, body mass index; C, cooperativeness; DSM-IV-TR, Diagnostic and Statistical Manual of Mental Disorders – fourth edition – text revisions; EDI-2, Eating Disorders inventory-2; HA, harm avoidance; MANQ, Meaning of Anorexia Nervosa Questionnaire; NS, novelty seeking; SD, self-directedness; TCI, Temperament and Character Inventory.