1. Introduction
Anorexia Nervosa (AN) is a serious psychiatric disorder characterized by persistent restriction of energy intake leading to significantly low body weight, an intense fear of gaining weight or of becoming fat, and a disturbance in the experience of body weight or shape. Diagnosing and assessing the severity of Anorexia Nervosa is a complex process, and for many years, clinicians have relied on various criteria to understand the depth and impact of this condition. One of the key indicators used in recent diagnostic frameworks, particularly within the DSM-5, is the Body Mass Index (BMI). BMI, a simple calculation of weight relative to height, has been adopted to provide specifiers for the severity of AN in adults. These BMI specifiers range from mild to extreme, aiming to categorize the level of physical health risk associated with different weight ranges in anorexia.
The introduction of BMI specifiers in 2013 was intended to bring more clarity to the diagnosis and treatment planning of AN. These specifiers include categories such as mild (BMI ≥ 17), moderate (BMI between 16 and 16.9), severe (BMI between 15 and 15.9), and extreme (BMI < 15). However, the validity and clinical utility of these BMI classifications, especially when applied to patients with very low BMIs requiring intensive inpatient treatment, have been increasingly questioned. While BMI is easily measurable and provides a quantitative metric of underweight status, its effectiveness in reflecting the broader psychopathology and predicting treatment outcomes in AN remains a topic of ongoing research and debate.
Previous studies, largely conducted with outpatients, have suggested that BMI specifiers may not consistently correlate with the severity of eating disorder psychopathology. Some research indicates that individuals with “extreme” AN, as defined by BMI, do not necessarily exhibit more severe eating disorder symptoms compared to those with higher BMIs. In some cases, they may even report less severe scores on measures of eating psychopathology. This raises concerns about whether BMI alone is sufficient to capture the multifaceted nature of AN severity, particularly in populations requiring hospitalization due to the acuteness of their condition.
Hospitalization for Anorexia Nervosa is often necessitated by dangerously low BMI or rapid weight loss, indicating a critical phase of the illness. Inpatient settings cater to patients who are in a very malnutrition-related acute phase, requiring intensive medical and psychological interventions. Within this specific group of patients, the application and relevance of BMI specifiers may differ from outpatient populations. Patients admitted to hospitals for AN frequently present with extreme BMIs, sometimes falling to critically low levels. This heterogeneity within the “extreme” BMI category itself – ranging from just under 15 to below 10 – suggests a potential limitation of the current BMI classification. It becomes questionable whether a single “extreme” category adequately differentiates the varying degrees of medical and psychological severity within this range.
This context highlights a critical gap in our understanding: Do current BMI specifiers effectively differentiate levels of severity among inpatients with anorexia nervosa, especially those with extremely low BMIs? Furthermore, is there a need to refine these specifiers, perhaps by introducing a “very extreme” category for patients with BMIs at the lowest end of the spectrum? Answering these questions is crucial for improving diagnostic accuracy, tailoring treatment approaches, and predicting outcomes for individuals hospitalized with severe Anorexia Nervosa.
To address these gaps, a study was designed to investigate the validity of BMI specifiers in the context of acute hospitalization for AN. The study aimed to:
- Evaluate the applicability of current DSM-5 BMI specifiers in inpatients with severe AN.
- Explore the potential utility of a “very extreme” BMI specifier (BMI ≤ 13.5) in this population.
- Assess whether BMI severity at admission predicts hospitalization outcomes for inpatients with AN.
This research endeavors to provide insights into the role of BMI in understanding and managing severe Anorexia Nervosa in inpatient settings, contributing to a more nuanced approach to Anorexia Bmi Diagnosis and severity assessment.
2. Materials and Methods
2.1. Participants
The study recruited participants from individuals seeking voluntary hospitalization at an Eating Disorders Center in Italy between December 2016 and July 2021. Potential participants were assessed for eligibility based on the following inclusion criteria:
- A primary diagnosis of Anorexia Nervosa, confirmed by an experienced psychiatrist using the Structured Clinical Interview for DSM-5 (SCID-5).
- Age 18 years or older.
- Absence of psychotic or bipolar disorders.
Initially, 193 individuals were considered for the study. However, after screening, 16 individuals were excluded due to incomplete assessments, and 9 declined to participate. The final sample comprised 168 inpatients diagnosed with Anorexia Nervosa. The study received ethical approval from the relevant institutional review board.
2.2. Treatment in Inpatient Setting
The treatment protocol in the inpatient setting was designed to address the acute medical and psychological needs of patients with severe AN. A multidisciplinary team delivered individualized treatment plans, in line with best practices for inpatient AN care. Notably, a significant majority (over 80%) of patients were admitted through the emergency room, indicating the acute and often life-threatening nature of their condition at admission. The primary focus of the inpatient intervention was to stabilize patients’ medical status, address immediate life-threatening conditions resulting from severe malnutrition, and enhance patient motivation for continued therapeutic engagement after discharge. Psychoeducation for families was also an integral component of the treatment approach.
2.3. Assessments and Measures
Upon admission to the hospital (T0) and at discharge (End of Treatment, EOT), comprehensive data were collected for each participant. This included:
- Sociodemographic and Clinical Characteristics: Gathered through clinical interviews, including age, sex, duration of illness, and history of previous hospitalizations.
- Body Mass Index (BMI): Measured by trained nurses at both admission and discharge, using standardized procedures for height and weight measurement.
- Eating Disorder Examination-Questionnaire (EDE-Q): A self-report questionnaire assessing eating disorder psychopathology, covering areas such as restraint, eating concern, shape concern, and weight concern. The EDE-Q provides a total score reflecting overall eating psychopathology severity.
- Eating Disorder Inventory-2 (EDI-2): Another self-report measure evaluating a range of psychological traits and symptoms relevant to eating disorders. The study focused on “core” subscales including Drive for Thinness, Bulimia, and Body Dissatisfaction.
- State-Trait Anxiety Inventory (STAI): Assessed both state anxiety (current anxiety level) and trait anxiety (general predisposition to anxiety).
- Beck Depression Inventory (BDI): A widely used self-report instrument measuring the severity of depressive symptoms.
- Body Shape Questionnaire (BSQ): Evaluated concerns and dissatisfaction with body shape.
- EQ-5D-VAS: A standardized instrument measuring health-related quality of life, including a Visual Analogue Scale (VAS) for self-rated health status and an index score.
All participants completed these assessments at both the beginning (T0) and end (EOT) of their inpatient treatment.
2.4. Statistical Data Analysis
The collected data were analyzed using SPSS statistical software. The statistical approach involved several steps to compare groups based on BMI severity and to assess changes during hospitalization.
- Comparison of Current BMI Specifiers: Patients were categorized into two groups based on the DSM-5 current BMI classification: Current Extreme Anorexia Nervosa (Current-E-AN, BMI < 15) and Non-Extreme Anorexia Nervosa (NE-AN, BMI ≥ 15). Independent sample t-tests were used to compare continuous variables (e.g., age, duration of illness, psychopathology scores), and Fisher’s exact tests were used for categorical variables (e.g., AN subtype). Cohen’s d was calculated to measure effect sizes for significant differences.
- Exploration of a “Very Extreme” BMI Specifier: To investigate a more granular classification at the lower end of the BMI spectrum, the Current-E-AN group was further divided. Three groups were formed: Very Extreme Anorexia Nervosa (VE-AN, BMI ≤ 13.5), Extreme Anorexia Nervosa (E-AN, BMI 13.6–14.9), and Non-Extreme Anorexia Nervosa (NE-AN, BMI ≥ 15). Analysis of Variance (ANOVA) was used to compare these three groups on continuous variables, with Bonferroni post-hoc tests for pairwise comparisons. Fisher’s exact tests were used for categorical variables. Effect sizes were measured using eta-squared (η2).
- Outcome Analysis: Repeated measures ANOVA was employed to examine changes in clinical outcomes (BMI, EDE-Q scores, EDI-2 subscales, anxiety, depression, body image, quality of life) across the VE-AN, E-AN, and NE-AN groups from admission to discharge. This analysis assessed the main effects of time (admission vs. discharge), group (BMI severity category), and the interaction between time and group (differential treatment response).
- Covariate Analysis: Given observed differences in length of hospitalization across BMI groups, Analysis of Covariance (ANCOVA) was used to control for the effect of hospitalization duration when comparing groups and assessing treatment outcomes. This step helped to isolate the specific effects of BMI severity from the influence of length of stay.
Statistical significance was set at p < 0.05 for all analyses. This rigorous statistical approach allowed for a comprehensive evaluation of BMI specifiers and their relationship to psychopathology and treatment outcomes in hospitalized patients with Anorexia Nervosa.
3. Results
3.1. Sociodemographic and Clinical Characteristics of the Sample
The study sample of 168 inpatients with Anorexia Nervosa was predominantly female (94.6%), reflecting the higher prevalence of AN in women. All participants were Caucasian and had voluntarily sought hospitalization. None of the patients left the treatment program against medical advice during their inpatient stay. The mean age of the sample was 24.3 years (SD = 9.5), with an age range from 18 to 56 years. The average duration of illness was 6.1 years (SD = 7.8), ranging from 0.5 to 40 years, indicating a sample with both relatively recent onset and chronic cases of AN. The mean BMI at admission was 14.3 (SD = 1.9), highlighting the significantly underweight status of this inpatient population.
Regarding AN subtypes, 74.4% of the sample were diagnosed with the restricting subtype (AN-R), and 25.6% with the binge-purging subtype (AN-BP). A substantial proportion of the sample (43.5%) reported previous hospitalizations for Anorexia Nervosa, indicating a history of severe or recurrent illness. The mean duration of the index hospitalization was 35.3 days (SD = 14.8), reflecting the intensity and length of inpatient treatment required for this patient group.
When comparing patients categorized as Current-E-AN (BMI < 15) versus NE-AN (BMI ≥ 15) based on current DSM-5 criteria, no significant differences were found in sex or history of previous hospitalizations. However, a significant difference emerged in AN subtypes: the Current-E-AN group had a significantly higher proportion of individuals diagnosed with the restricting subtype (AN-R) compared to the NE-AN group (72.8% vs. 27.2%, p = 0.003).
Comparing the three BMI severity groups – VE-AN (BMI ≤ 13.5), E-AN (BMI 13.6–14.9), and NE-AN (BMI ≥ 15) – revealed that the VE-AN group also had a significantly higher proportion of patients with the restricting AN subtype (AN-R) when compared to both the E-AN and NE-AN groups (44% vs. 28.8% and 27.2% respectively, p = 0.001). No significant differences were observed across the three BMI severity groups in terms of sex distribution or history of previous hospitalizations.
3.2. Comparison Between Groups Based on Current BMI Classification
Analyzing the differences between the Current-E-AN and NE-AN groups based on the current DSM-5 BMI classification revealed several significant findings (Table 1). While there were no differences in age or duration of illness between the groups, a statistically significant difference was found in the length of hospitalization. The Current-E-AN group had a significantly longer mean duration of hospitalization (37.5 days) compared to the NE-AN group (30.3 days, p = 0.009, Cohen’s d = 0.43).
Contrary to expectations, the NE-AN group (BMI ≥ 15) reported significantly more severe psychopathology across multiple measures compared to the Current-E-AN group (BMI < 15). Specifically, the NE-AN group scored higher (indicating greater severity) on all subscales of the EDE-Q (Restraint, Eating Concern, Shape Concern, Weight Concern, and Total Score), all core subscales of the EDI-2 (Drive for Thinness, Bulimia, Body Dissatisfaction), both State and Trait Anxiety (STAI), Beck Depression Inventory (BDI), and Body Shape Questionnaire (BSQ). Effect sizes for these differences ranged from medium to large, suggesting clinically meaningful distinctions. Notably, quality of life, as measured by the EQ-5D-VAS and Index, did not differ significantly between the Current-E-AN and NE-AN groups.
3.3. Comparison Across Groups Based on Novel Exploratory BMI Classification
The comparison across the three BMI severity groups (VE-AN, E-AN, NE-AN) based on the novel exploratory classification yielded further insights (Table 2). Similar to the previous analysis, no significant differences emerged in age or duration of illness across the three groups. However, a significant overall difference in length of hospitalization was observed (p = 0.004, η2 = 0.06). Post-hoc analyses revealed that while the VE-AN and E-AN groups had comparable lengths of stay, and the E-AN and NE-AN groups also had similar durations, the VE-AN group had a significantly longer hospitalization compared to the NE-AN group.
Regarding psychopathology, the VE-AN and E-AN groups exhibited similar scores across most measures. However, both the VE-AN and E-AN groups reported significantly less severe eating psychopathology, anxiety, depressive symptoms, and body image concerns compared to the NE-AN group. Specifically, the NE-AN group had significantly higher scores on all EDE-Q subscales and total score, EDI-2 Drive for Thinness, Bulimia, and Body Dissatisfaction, STAI State and Trait Anxiety, BDI, and BSQ compared to the VE-AN and E-AN groups. Effect sizes for these differences ranged from medium to large (η2 = 0.07 to 0.17). Again, similar to the comparison of Current-E-AN and NE-AN, no significant differences were found in quality of life (EQ-5D-VAS and Index) across the three BMI severity groups.
3.4. Clinical Outcome Across BMI Severity Groups During Hospitalization
Analysis of clinical outcomes during hospitalization across the VE-AN, E-AN, and NE-AN groups revealed that all groups demonstrated significant improvement on most measures from admission (T0) to discharge (EOT) (Table 3). This included significant improvements in BMI, all EDE-Q subscales and total score, EDI-2 Bulimia, STAI State and Trait Anxiety, BDI, BSQ, and EQ-5D-VAS and Index. However, body dissatisfaction, as measured by the EDI-2 Body Dissatisfaction subscale, did not show significant improvement during hospitalization across any of the groups.
Consistent with baseline comparisons, significant main effects of group were observed for most measures, indicating that the groups differed in their overall scores throughout hospitalization. However, importantly, no significant timegroup interactions were found for the majority of outcomes, suggesting that all three BMI severity groups showed similar trajectories of improvement during hospitalization, despite differing baseline severity levels. The only significant timegroup interaction was observed for BMI change during hospitalization, indicating that the VE-AN group exhibited the greatest improvement in BMI from admission to discharge compared to the E-AN and NE-AN groups.
Given the significant differences in length of hospitalization across the BMI groups, further analyses controlled for hospitalization duration. After statistically controlling for length of stay, the significant improvements in EDI-2 Drive for Thinness, Trait Anxiety, Body Image Concerns (BSQ), and Quality of Life (EQ-5D-VAS and Index) over time were no longer significant. This suggests that the observed improvements in these specific measures may be time-dependent and influenced by the duration of hospitalization.
4. Discussion
This study aimed to evaluate the validity of current and potential novel BMI specifiers for Anorexia Nervosa in an inpatient setting, focusing on patients requiring acute hospitalization. The findings yielded three key insights:
Firstly, when comparing patients with extreme and non-extreme AN based on current DSM-5 criteria, the non-extreme AN group (BMI ≥ 15) exhibited comparable quality of life but paradoxically reported significantly more severe eating and general psychopathology, along with greater body image concerns compared to the extreme AN group (BMI < 15). This finding challenges the straightforward assumption that lower BMI directly equates to greater overall psychopathology in hospitalized AN patients.
Secondly, exploring a “very extreme” AN specifier (BMI ≤ 13.5) revealed a similar trend. The very extreme AN group, despite having the lowest BMIs and requiring the longest hospital stays, reported similar levels of psychopathology to the extreme AN group (BMI 13.6–14.9), but both of these groups reported less severe psychopathology, anxiety, depressive symptoms, and body image concerns compared to the non-extreme AN group (BMI ≥ 15). This suggests that further subdividing the “extreme” BMI category may not effectively differentiate levels of psychopathology in this inpatient population.
Thirdly, despite baseline differences in psychopathology severity, all three BMI groups (VE-AN, E-AN, NE-AN) showed comparable positive responses to acute inpatient treatment, with similar trajectories of improvement across most measures. The length of hospitalization, however, appeared to significantly influence clinical outcome, particularly for measures like drive for thinness, trait anxiety, body image concerns, and quality of life, suggesting that improvement in these areas may require more extended treatment duration.
These findings are somewhat contrary to the intuitive expectation that lower BMI would consistently correlate with greater psychopathology in Anorexia Nervosa. The observation that patients with non-extreme BMI (BMI ≥ 15) reported more severe eating disorder symptoms, anxiety, and depression than those with extreme or very extreme BMIs is noteworthy. One potential explanation for this unexpected result is the phenomenon of “ego-syntonicity” in AN. Anorexia Nervosa is often described as an ego-syntonic disorder, meaning that the symptoms and behaviors are often perceived by the individual as aligned with their values and self-image. It is possible that patients with very extreme BMI, being in a more physically compromised state, may paradoxically minimize their psychological distress or lack awareness of the severity of their condition as a defense mechanism or due to altered cognitive processing associated with severe malnutrition. In contrast, patients with slightly higher BMIs, while still meeting diagnostic criteria for AN, may have greater insight into their psychopathology and thus report higher levels of distress and symptom severity on self-report measures.
The finding that length of stay significantly impacted clinical outcomes, particularly for drive for thinness, trait anxiety, body image, and quality of life, underscores the importance of adequate treatment duration in inpatient settings. These aspects of psychopathology and well-being may require more time to improve within the structured environment of inpatient care. The lack of significant time*group interactions for most outcomes suggests that BMI severity at admission may not be a strong predictor of differential treatment response in terms of psychopathology reduction during hospitalization, although patients with very extreme BMI did show greater BMI gain during their stay, as expected.
The study’s findings raise questions about the clinical utility of BMI specifiers alone in reflecting the full spectrum of severity in hospitalized Anorexia Nervosa patients. While BMI is undoubtedly a critical indicator of physical health risk and medical instability, it may not fully capture the complexity of psychopathology and subjective distress experienced by individuals with AN, particularly in acute inpatient settings. The results suggest that a more comprehensive assessment of severity in AN should consider not only BMI but also the individual’s level of insight, psychological symptoms, functional impairment, and perhaps clinician-rated severity measures in addition to self-report questionnaires.
Limitations of this study include its focus on inpatients with severe AN, which may limit the generalizability of findings to outpatient populations or those with less severe forms of the disorder. Additionally, the study relied primarily on self-report measures, which may be subject to biases, particularly in individuals with severe eating disorders who may have distorted self-perception or limited insight. Future research could benefit from incorporating clinician-rated measures of severity and exploring the longitudinal course of psychopathology and outcomes beyond the inpatient setting.
5. Conclusions
In conclusion, the findings of this study do not provide strong support for the validity of current DSM-5 BMI specifiers as robust indicators of psychopathology severity in inpatients with Anorexia Nervosa. Furthermore, the exploration of a “very extreme” BMI specifier (BMI ≤ 13.5) did not reveal significant differences in psychopathology compared to the current “extreme” category. Paradoxically, patients with non-extreme BMI (BMI ≥ 15) reported more severe eating disorder symptoms and psychological distress than those with lower BMIs, challenging the assumption that BMI directly reflects psychopathological severity in this population.
While BMI remains a crucial medical indicator in Anorexia Nervosa, particularly for assessing physical risk and monitoring weight restoration, these results suggest that relying solely on BMI specifiers may be insufficient for comprehensively understanding and classifying the severity of AN, especially in inpatient settings. A more holistic approach to anorexia bmi diagnosis and severity assessment, incorporating multiple dimensions of psychopathology, insight, and functional status, may be necessary to guide individualized treatment planning and predict outcomes more effectively. Future research should continue to explore more nuanced and comprehensive methods for assessing severity in Anorexia Nervosa, moving beyond a singular focus on BMI to encompass the multifaceted nature of this complex disorder.
Author Contributions: Conceptualization, E.M. and G.A.-D.; software, P.L. and M.M.; methodology, E.M. and M.M.; formal analysis, F.T. and E.M.; data curation, F.T., E.M. and P.L.; writing—original draft preparation, F.T. and E.M.; writing—review and editing, G.A.-D., P.L. and M.M.; supervision, G.A.-D.; project administration, E.M. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: The study was conducted in accordance with the Declaration of Helsinki, and approved by the Ethics Committee) of the “Città della Salute e della Scienza” hospital at the University of Turin, Italy, with protocol number 0073951 (date of approval 9 July 2021).
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.
Data Availability Statement: Anonymized data can be required to the last author upon reasonable request.
Conflicts of Interest: The authors declare no conflict of interest.
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References: (Same as original article – omitted for brevity)
Associated Data: Data Availability Statement: Anonymized data can be required to the last author upon reasonable request.
