Baseline phase angle is associated with morphofunctional status, length of stay, and hospitalization costs in severe anorexia nervosa: the EDIGRA study

Abstract

Background

Phase angle (PhA), derived from bioelectrical impedance analysis, reflects bioelectrical properties related to cellular mass and fluid distribution and has been proposed as a marker of malnutrition severity. However, its exploratory association with morphofunctional status and hospitalization-related outcomes in severe anorexia nervosa (AN) remains insufficiently characterized. This study aimed to explore whether baseline PhA is associated with morphofunctional status, length of stay, and inpatient costs in a specialized eating disorders unit.

Methods

In this prospective cohort study, 42 female inpatients with severe AN or other specified feeding and eating disorder were assessed at admission. Patients were stratified into tertiles according to PhA. Anthropometry, body composition by bioelectrical impedance vector analysis (BIVA), intracellular and extracellular water distribution, handgrip strength, muscle and abdominal ultrasound parameters, biochemical markers, and length of stay were recorded. Hospitalization costs were estimated using standardized diagnosis-related group daily expenditure.

Results

Mean PhA values were 4.0°, 4.7°, and 5.5° in the low, mid, and high tertiles, respectively. Importantly, total body weight did not differ significantly across tertiles. In contrast, body cell mass index increased progressively (5.2, 6.1, and 6.9 kg/m²). Higher PhA was associated with greater rectus femoris cross-sectional area (2.3 vs. 3.7 cm²) and higher handgrip strength (20.3 vs. 24.9 kg), consistent with more favorable muscle structure and function in unadjusted comparisons. Hydration profiles also differed: extracellular water proportion decreased (52.7% to 40.3%), while intracellular water increased (44.1% to 53.3%) across tertiles. Median length of stay declined from 58.4 to 41.3 days, with corresponding reductions in estimated hospitalization costs (€36,523 to €25,829), which should be interpreted descriptively because cost estimates were largely driven by hospitalization duration. ROC analysis showed modest discriminatory performance for prolonged hospitalization (AUC = 0.65; exploratory threshold: 4.5–4.6°).

Conclusions

Baseline PhA was associated with differences in morphofunctional profiles and hospitalization trajectories despite similar body weight, suggesting that it may capture morphofunctional variability not reflected by anthropometry alone. Its associations with cellular mass, hydration distribution, muscle function, and length of stay suggest that PhA may provide complementary descriptive information within a broader clinical assessment. However, given the modest discriminative performance, sample size, and lack of multivariable adjustment, PhA should not be considered a standalone prognostic tool. Larger multicenter studies are needed to validate thresholds and formally test whether PhA provides incremental information beyond conventional anthropometric, clinical, psychological, and organizational factors.

Ethics approval

Approved by the Provincial Research Ethics Committee of Granada (SICEIA-2024-003069).

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s40337-026-01687-7.

Plain language summary

People with severe anorexia nervosa often need hospital treatment to restore their physical health safely. Healthcare professionals usually monitor recovery using body weight and body mass index (BMI), but these measures do not always reflect how well the body’s tissues, muscles, and cells are functioning. This study examined whether a measurement called phase angle could provide additional and more detailed information about the physical condition of the body. Phase angle is obtained from a painless body composition test that uses a very small electrical current. It gives information about cell mass, muscle condition, and body fluid balance. In particular, it reflects differences between water inside the cells and water outside the cells, which may be altered in severe malnutrition. Higher phase angle values are generally associated with more favorable cellular and muscle health. We studied 42 women admitted to hospital with severe anorexia nervosa or other specified feeding and eating disorder. Although patients had similar body weight, those with higher phase angle values showed stronger muscles, more favorable tissue-related measurements, and better body fluid balance at admission. They also had shorter hospital stays and lower treatment costs, although these lower costs were largely explained by shorter hospital stays. In contrast, patients with lower phase angle values tended to require longer inpatient care. These findings suggest that phase angle may help healthcare teams better understand the physical condition of patients beyond weight alone. Including phase angle in routine assessment may provide complementary information for clinical monitoring, but it should be used alongside broader medical, psychological, and organizational assessment during hospital treatment for anorexia nervosa, rather than as a standalone tool for predicting recovery, hospital stay, or discharge timing.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s40337-026-01687-7.

Introduction

Anorexia nervosa (AN) is a severe psychiatric disorder characterized by self-induced weight loss, distorted body image, and an intense fear of weight gain, resulting in profound malnutrition and multisystem involvement [1]. It is associated with the highest mortality rate among psychiatric illnesses, largely due to medical complications of starvation and suicide [2–4]. From an epidemiological perspective, AN is currently the third most common chronic disease among female adolescents, after asthma and obesity [5, 6]. Despite advances in psychological and nutritional interventions, physical recovery remains difficult to monitor objectively, particularly during inpatient nutritional rehabilitation [7]. In this setting, patients with comparable body weight or BMI trajectories may exhibit markedly different degrees of cellular integrity, muscle depletion, and functional impairment, complicating the descriptive assessment of physiological recovery within broader clinical evaluation [8].

Traditionally, body weight and BMI have been used as the primary indicators of nutritional status and treatment response in AN and remain central to diagnostic severity classifications, despite increasing recognition of their limitations in reflecting overall nutritional and functional recovery [9] However, BMI is a static and indirect marker that does not adequately reflect alterations in body composition or cellular health. In severely malnourished patients, weight changes may occur as a result of extracellular fluid shifts or preferential fat deposition, without parallel restoration of lean mass or metabolically active tissue [8, 10–12]. Consequently, reliance on BMI alone may lead to an overestimation of true physiological recovery and fails to capture clinically relevant changes in nutritional quality. This limitation is particularly relevant in inpatient settings, where early refeeding is frequently accompanied by rapid weight gain that may not correspond to meaningful functional or cellular recovery [13].

To overcome these limitations, interest has grown in morphofunctional assessment, an integrated approach combining structural and functional markers of nutritional status [14, 15]. Within this framework, bioelectrical impedance analysis (BIA) and bioelectrical impedance vector analysis (BIVA) have emerged as practical bedside tools [16, 17]. BIA evaluates the electrical properties of body tissues by measuring resistance and reactance, which are influenced by hydration status, tissue composition, and cell membrane characteristics [12]. From these parameters, PhA is derived as the arctangent of the ratio between reactance and resistance and represents the delay (phase shift) between voltage and current as the electrical signal passes through body tissues [18]. Physiologically, PhA reflects the capacitive properties of cell membranes and the quantity and quality of metabolically active tissue, commonly referred to as body cell mass [19, 20]. Intact, healthy cell membranes act as effective capacitors, increasing reactance and resulting in higher PhA values, whereas membrane damage, reduced cell mass, or cellular atrophy lead to lower reactance and reduced PhA [19, 20]. Low PhA values are consistently associated with cellular depletion, impaired membrane function, and poor nutritional status, whereas higher values indicate better cellular integrity and metabolic reserve [21–24]. In this context, PhA has been proposed as a complementary marker capable of capturing qualitative aspects of nutritional status that are not reflected by body weight or BMI alone [25]. In several clinical settings, including oncology, chronic organ failure, critical illness, and inflammatory diseases, PhA has also been associated with adverse clinical outcomes [26–31]. In the context of AN, several studies have shown that PhA is significantly reduced compared with healthy controls and constitutionally lean individuals, suggesting that it reflects qualitative alterations in body composition rather than low body weight alone [32, 33]. Moreover, PhA has been shown to change dynamically during nutritional rehabilitation and to correlate with body cell mass and metabolic parameters, supporting its potential role as a marker of physical recovery [17, 23, 34].

In parallel, complementary morphofunctional techniques provide additional insight into tissue-level and functional recovery. Muscle ultrasound (MU) allows direct, non-invasive assessment of muscle and adipose tissue compartments, including muscle thickness, cross-sectional area, and subcutaneous fat distribution [35–37]. Reduced muscle size assessed by ultrasound reflects severe muscle wasting, while increases during treatment indicate true lean tissue restoration [8, 38, 39]. In the same way, handgrip strength (HGS) is a simple and reliable measure of muscle function and overall functional capacity, with low values associated with malnutrition, sarcopenia, and adverse clinical outcomes [40–42]. Improvements in HGS during refeeding reflect functional recovery of the neuromuscular system and correlate with changes in lean mass [17, 43].

Although BIA-derived PhA, MU, and HGS have each been separately investigated in AN, their combined application as part of a comprehensive morphofunctional assessment remains limited, and data linking these measures to clinically relevant outcomes such as length of hospital stay and healthcare costs are null. Moreover, few studies have examined whether PhA provides complementary information beyond traditional severity markers or whether exploratory PhA thresholds may be generated for future validation in inpatient care. Importantly, few studies have evaluated these tools as dynamic markers of recovery rather than static descriptors of disease severity [17, 23, 32, 33].

Therefore, the aim of this study was to explore whether baseline phase angle is associated with morphofunctional status and inpatient outcomes in patients with severe anorexia nervosa. Specifically, we evaluated its association with body composition assessed by BIVA and muscle and abdominal ultrasound, muscle function measured by handgrip strength, hydration distribution parameters, and clinically relevant outcomes including length of hospitalization and healthcare costs. By integrating structural, functional, and economic indicators, this study sought to assess the exploratory relevance of phase angle as a complementary descriptive variable alongside conventional weight-based measures in inpatient AN care.

Methods

Study and participants

The study enrolled 42 female patients (mean age 28.7 ± 13.5 years) admitted to the Eating Disorders Hospitalization Unit (EDHU) at Virgen de las Nieves University Hospital between 2020 and 2023. All had a DSM-5 diagnosis of restrictive AN (n = 41) or other specified feeding and eating disorder (OSFED; n = 1) with a restrictive pattern prior to admission [44]. The inclusion of the single OSFED case was based on its restrictive phenotype, which closely matched the clinical presentation and admission criteria of the rest of the cohort, ensuring phenotypic homogeneity.

Eligibility criteria included age ≥ 16 years and a confirmed diagnosis of AN or OSFED. During hospitalization at the EDHU, patients received the standard clinical care protocol, which included psychiatric and nutritional management aimed at normalizing eating behaviors, improving illness insight, and supporting behavioral change for safe outpatient transition. Individualized diets were prescribed and medically supervised.

To minimize potential distress and reduce psychological burden, patients were blinded to the results of all morphofunctional measurements obtained during hospitalization. The study protocol was approved by the Provincial Research Ethics Committee of Granada (SICEIA-2024-003069).

Classical anthropometric measurements

At admission, height was measured with a stadiometer, and body weight was obtained using a calibrated scale (SECA 665, Hamburg, Germany) with certified test weights (accuracy ± 0.1 kg). Anthropometric assessment included calf and arm circumferences and triceps skinfold thickness, performed following standardized protocols [45]. All measurements were conducted by trained professionals under controlled conditions to ensure reliability and minimize variability.

Bioelectrical impedance vector analysis

BIVA was used to evaluate indirect cellular integrity and hydration-related body composition through resistance–reactance vector analysis, providing morphofunctional information complementary to weight-based anthropometric indices [46]. BIVA was performed with a 50 kHz phase-sensitive analyzer (BIA 101 AKERN, Pontassieve, Italy) using tetrapolar 800 mA electrodes on the right hand and foot. The method applied an alternating current to measure PhA from resistance (R) and reactance (X_C). All measurements were performed with the patient in a supine position on a stretcher, without jewelry or metallic objects, and either immediately after breakfast or at least two hours after the breakfast, in accordance with manufacturer and guideline recommendations to ensure measurement stability [12, 47, 48]. Electrodes were placed homolaterally on the right side, following standard guidelines: in the upper limb, centered in the proximal electrode area between the distal prominences of the radius and ulna, and in the lower limb, between the medial and lateral malleoli of the ankle, ensuring consistent anatomical positioning [17]. Data were interpreted using resistance-reactance (R-X_C) graphs using bioelectrical impedance vector references of Campa et al. [49]. Analyzed parameters included body electrical properties, fat and muscle compartments, cellular mass, and hydration status and distribution. The device was calibrated once per week using a dedicated tester with a circuit of known R and X_C, in accordance with manufacturer specifications. All measurements were performed under standardized conditions by trained professionals to ensure reproducibility and minimize variability.

Muscle and abdominal ultrasound

MU was used as a non-invasive bedside tool to assess muscle mass, architecture, and regional adipose tissue distribution, providing objective structural information on nutritional status beyond conventional anthropometric measures [35]. Muscle ultrasound was performed with a HITACHI ALOKA F37 scanner (Hitachi Healthcare, Tokyo, Japan) using a 10–12 MHz linear transducer (Aloka UST-5413), following standardized protocols [17]. Patients were assessed in a supine position with proper limb alignment, and water-based gel was applied to optimize image quality. For muscle evaluation, the right rectus femoris (RF) was examined at a fixed point one-third of the distance between the patella and the iliac crest, obtaining the cross-sectional area (RF-CSA), its X- and Y-axis, and leg subcutaneous fat (L-SAT). At the abdominal level, measurements were taken midway between the xiphoid process and the umbilicus to determine total subcutaneous abdominal adipose tissue (T-SAT), superficial subcutaneous fat (S-SAT), and visceral adipose tissue (VAT), providing an estimate of fat distribution and accumulation. Abdominal measurements were obtained under relaxed breathing conditions to minimize variability. As a quality control procedure, three images were acquired for each measurement, and the mean value was used for analysis. All assessments were conducted by trained operators using standardized acquisition protocols, with careful control of probe pressure to avoid tissue compression. Inter-observer variability was controlled, with a maximum accepted variability of 5%. While these procedures aimed to enhance reproducibility, some degree of variability inherent to ultrasound-based measurements cannot be completely excluded.

Handgrip strength analysis

HGS was assessed as a functional indicator of muscle performance and global physical status, complementing the structural information provided by body composition and ultrasound measurements. HGS of the dominant hand was measured using a Jamar dynamometer (Asimow Engineering Co., Los Angeles, CA, USA). Patients were seated with the shoulder adducted, elbow flexed at 90°, and wrist and forearm in a neutral position [17]. They performed three maximal voluntary contractions, each separated by a one-minute rest, and the mean value was recorded according to standardized protocols. All measurements were carried out by trained professionals under controlled conditions to ensure reliability and minimize variability.

Biochemical analysis

Biochemical parameters were assessed to evaluate nutritional and inflammatory status, including markers of protein metabolism (total protein, albumin, prealbumin, urea, and creatinine), electrolytes and minerals (sodium, potassium, magnesium, phosphorus, calcium, and iron), lipid profile (total cholesterol and triglycerides), vitamin status (folic acid and vitamin B12), inflammation (C-reactive protein), glucose and thyroid-stimulating hormone (TSH).

Clinical outcomes: length of stay and costs

The cost of eating disorder–related hospitalizations within the mental health setting was estimated using data from the official diagnosis-related group (DRG) system. Specifically, the DRG corresponding to “Trastorno de la conducta alimentaria” has a standardized cost of €625 per inpatient day. DRGs are case-mix classification systems designed to estimate the average healthcare resource utilization associated with a given diagnostic category, including inpatient bed occupancy, medical and nursing care, diagnostic procedures, pharmacological treatment, nutritional support, and infrastructure-related costs [50]. This standardized daily cost was multiplied by the individual length of hospital stay to estimate total hospitalization costs for each patient, according to the formula: Individual hospitalization cost = DRG daily cost × length of stay. This methodology provides a uniform and conservative estimate of inpatient costs, while acknowledging that it may not fully capture additional DRGs, comorbid conditions, or case-specific interventions requiring supplementary resources.

Subgroup elaboration and statistical analysis

The formal analysis was performed using IBM SPSS 25 (Chicago, IL, USA). The 42 patients were stratified into three subgroups according to PhA tertiles, which were used as an exploratory framework to describe progressively different bioelectrical profiles related to cellular integrity. The use of tertiles was chosen to facilitate descriptive interpretability, to allow comparison of extreme bioelectrical profiles, and to explore potential non-linear associations between PhA and morphofunctional, clinical, and economic outcomes. This approach also enabled an exploratory comparison of patients with potentially less favorable recovery-related profiles, without implying independent risk stratification or formally demonstrated clinical utility. Accordingly, the low-PhA group (< 4.4°) included 13 patients, the mid-PhA group (4.4–5.1°) comprised 16 patients, and the high-PhA group (> 5.1°) included 13 patients.

Normality and homogeneity of variances were evaluated using the Shapiro–Wilk and Levene’s tests, respectively. For variables satisfying assumptions of normal distribution and homoscedasticity, group differences were analyzed using one-way analysis of variance (ANOVA) followed by Tukey’s honestly significant difference (HSD) post hoc test. When the assumption of homogeneity of variances was not met, one-way ANOVA with Games–Howell post hoc adjustment was employed. Both post hoc procedures control the family-wise error rate, maintaining the overall type I error at α = 0.05.

To evaluate differences in hospitalization dynamics according to PhA status, time-to-discharge analyses were conducted. Kaplan–Meier survival curves were generated for each PhA tertile and compared using the log-rank test. Complementary Breslow and Tarone–Ware tests were also applied to explore potential differences in early versus late discharge patterns across groups, providing an additional descriptive assessment of hospitalization trajectories. Statistical significance was defined as p < 0.05.

To assess the exploratory discriminatory performance of PhA in relation to prolonged hospitalization, receiver operating characteristic (ROC) curve analyses were performed. Prolonged hospital stay was primarily defined as a length of stay above the cohort median (45 days), with an additional sensitivity analysis using the 75th percentile (56 days) to examine the consistency of the exploratory findings. Exploratory PhA cut-off values were determined using Youden’s index.

In addition, Partial Least Squares Discriminant Analysis (PLS-DA) was conducted using MetaboAnalyst version 5.0 to explore multivariate morphofunctional patterns associated with PhA tertiles. Given the exploratory nature of this analysis and the modest sample size, PLS-DA was used as a hypothesis-generating tool to characterize global differences in nutritional, body composition, and functional profiles rather than for confirmatory inference. Prior to analysis, variables were normalized and auto-scaled to ensure comparability across domains. Variables with a variable importance in projection (VIP) score > 1 were considered the main contributors to the descriptive separation between groups, while recognizing that several variables originated from the same bioimpedance framework and may therefore be mathematically or methodologically interdependent. An intention-to-treat approach was applied, including two patients who did not complete the full therapeutic program, thereby preserving the integrity of the original cohort and reducing the potential influence of attrition.

Results

Anthropometric and functional assessment

Patients were stratified into tertiles according to PhA (Table 1), and results are presented as mean values with 95% confidence intervals to provide an indication of estimate precision. Across these tertiles, no significant differences in body weight were observed, with mean values ranging from 35.3 kg in the low PhA tertile to 39.4 kg in the high PhA tertile. Pairwise comparisons between tertiles, including mean differences and exact p-values, are reported in Table S1, with post hoc tests selected according to variance homogeneity assumptions. Given the large number of comparisons performed, the potential for type I error inflation should be acknowledged; therefore, results are interpreted with caution, emphasizing the magnitude, direction, and consistency of differences rather than isolated p-values. These descriptive findings suggest that patients with comparable body weight may nonetheless exhibit differences in bioelectrical characteristics.

Table 1.

Population characteristics stratified by phase angle (low, medium, and high)

	Low PhA		Mid PhA	High PhA
Anthropometry
Weight (kg)	35.3 (32.2–38.5)a		37.4 (35.2–39.7) a	39.4 (37.2–41.6) a
BMI (kg/m2)	13.5 (12.5–14.5)a		14.5 (13.7–15.2) ab	15.0 (14.4–15.7) b
AC (cm)	16.7 (15.2–18.7)a		18.3 (16.8–20.3) ab	19.0 (17.5–21.0) b
CC (cm)	26.8 (24.8–29.3)a		27.2 (25.2–29.7) a	29.1 (27.1–31.6) a
TST (mm)	2.8 (1.8–4.3)a		4.5 (3.5-6.0) ab	5.4 (4.4–6.9) b
BIVA
PhA (°)	4.0 (3.8–4.2)a		4.7 (4.6–4.8) b	5.5 (5.3–5.7) c
R (Ω)	705.8 (634.0-777.7)a		709.4 (669.4-749.4) a	754.2 (706.4–802.0) a
XC (Ω)	49.2 (43.4–54.9)a		58.6 (55.2–62.3) b	73.2 (67.3–79.0) c
BCM (kg)	13.6 (12.2–15.0)a		15.8 (15.0-16.6) b	18.0 (16.9–19.2) c
BCMI (kg/m²)	5.2 (4.8–5.6)a		6.1 (5.8–6.5) ab	6.9 (6.5–7.2) b
FM (kg)	2.5 (1.9–3.1)a		3.6 (2.5–4.7) a	4.3 (2.8–5.8) a
FFM (kg)	32.8 (30.0-35.6)a		33.9 (32.2–35.5) a	35.1 (32.9–37.3) a
TMM (kg)	18.0 (16.1–19.9)a		17.9 (16.7–19.1) a	18.0 (16.2–19.8) a
ASMM (kg)	12.0 (10.8–13.2)a		12.3 (11.6–13.0) a	13.1 (12.0-14.1) a
SMI (kg/m²)	6.7 (6.2–7.2)a		6.9 (6.4–7.4) a	6.8 (6.4–7.2) a
Hydration (%)	72.9 (71.6–74.3)a		72.8 (72.0-73.6) a	71.1 (69.5–72.7) a
TBW (%)	74.7 (71.1–78.3)a		70.7 (67.8–73.6) ab	65.9 (63.0-68.9) b
ECW (%)	52.7 (42.8–62.5)a		42.5 (31.0–54.0) a	40.3 (29.2–51.4) a
ICW (%)	44.1 (41.0-47.2)a		49.4 (47.1–51.7) b	53.3 (51.4–55.3) c
BM (kcal/day)	1144.7 (1103.2-1186.2)a		1208.3 (1183.9-1232.7) b	1272.0 (1238.8-1305.1) c
Nutrition	407.7 (370.9-444.6)a		477.8 (445.0-510.5) b	544.1 (518.4-569.8) c
Functional Measurement
HGS max (kg)	20.3 (17.3–24.3)a		19.0 (16.0–23.0) a	24.9 (21.9–28.9) b
Muscle Ultrasound
RF-CSA (cm²)	2.3 (1.1–3.5)a		3.6 (2.6–4.6) ab	3.7 (3.3–4.0) b
RF-X-axis (cm)	3.0 (2.4–3.6)a		3.4 (2.7–4.1) a	3.7 (3.3–4.1) b
RF-Y-axis (cm)	0.9 (0.7–1.1)a		1.4 (0.9–1.9) ab	1.2 (1.1–1.4) b
L-SAT (cm)	0.6 (0.1–1.3)a		0.5 (0.4–1.4) a	0.6 (0.3–0.8) a
Abdominal Ultrasound
T-SAT (cm)	0.3 (0.1–0.6)a		0.3 (0.1–0.6) a	0.8 (0.5–1.1) b
S-SAT (cm)	0.15 (0.02–0.32)a		0.12 (0.00–0.23) a	0.43 (0.2–0.7) a
VAT (cm)	0.25 (0.06–0.56)a		0.30 (0.17–0.77) a	0.29 (0.1–0.4) a
Biochemical Analysis
Glucose (mg/dL)	65.5 (57.8–73.2)a		69.0 (63.9–74.1) a	74.8 (68.8–80.8) a
Urea (mg/dL)	35.3 (6.3–64.3)a		22.9 (5.1–40.7) a	22.0 (3.4–47.4) a
Creatinine (mg/dL)	0.66 (0.56–0.77)a		0.70 (0.62–0.78) a	0.69 (0.58–0.79) a
Total Protein (g/dL)	6.6 (6.0-7.2)a		6.5 (5.9-7.0) a	6.9 (6.2–7.6) a
Albumin (g/dL)	4.2 (3.7–4.6)a		4.2 (4.0-4.5) a	4.5 (4.1-5.0) a
Prealbumin (mg/dL)	31.3 (17.9–44.6)a		24.2 (21.3–27.1) a	40.0 (10.4–69.6) a
Sodium (mEq/L)	129.5 (109.5-149.5)a		140.1 (138.9-141.2) a	139.6 (138.1–141.0) a
Potassium (mEq/L)	4.1 (3.9–4.3)a		4.2 (4.0-4.3) a	4.2 (4.0-4.5) a
Na/K (Ratio)	1.85 (1.70–2.01)a		1.50 (1.42–1.59) b	1.23 (1.13–1.33) c
Magnesium (mg/dL)	2.0 (1.9–2.2)a		2.0 (1.9–2.2) a	2.1 (1.9–2.2) a
Phosphorus (mg/dL)	3.4 (3.0-3.9)a		3.7 (3.3–4.2) a	3.7 (3.4–3.9) a
Calcium (mg/dL)	9.0 (8.6–9.4)a		8.8 (8.4–9.2) a	9.3 (8.9–9.7) a
Iron (µg/dL)	95.7 (67.3–124.0)a		71.0 (51.6–90.4) a	86.8 (19.9-153.7) a
C-reactive Protein (mg/L)	3.0 (1.0–7.0)a		0.4 (0.2–0.5) a	12.4 (1.1–23.6) a
Total Cholesterol (mg/dL)	183.0 (153.1-212.9)a		189.1 (144.3–234.0) a	101.4 (25.1-177.8) a
Triglycerides (mg/dL)	107.2 (80.9-133.4)a		80.1 (58.3-101.8) a	114.9 (58.9-170.9) a
Folic Acid (ng/mL)	16.3 (2.4–34.9)a		7.1 (4.1–10.0) a	44.0 (10.6–77.4) b
Vitamin B12 (pg/mL)	695.3 (477.2-913.4)a		701.2 (333.8-1068.6) a	446.0 (298.6-593.4) a
TSH (µU/mL)	1.67 (0.78–2.55)a		1.52 (0.91–2.14) a	4.39 (4.3–13.1) a
Hospitalization Data
Hospital Stay (days)	58.4 (45.4–71.4)a		44.1 (36.9–51.2) ab	41.3 (31.7–51.0) b
Mean Hospitalization Cost (€)	36523.4 (28393.2-44653.5)a		27580.14 (23077.3-32020.5) ab	25829.02 (19825.2-31895.4) b

AC Arm circumference, ASMM Appendicular skeletal muscle mass, BCM Body cell mass, MB Basal metabolism, BMI Body mass index, CC C alf circumference, CI Confidence interval, ECW Extracellular water, FFM Fat-free mass, FM Fat mass, HGS max Maximum handgrip strength, PhA Phase angle, RF-CSA Rectus femoris cross-sectional area, RF-X-axis Rectus femoris X-axis length, RF-Y-axis Rectus femoris Y-axis length, R Resistance, SAT Subcutaneous adipose tissue of leg (L), and superficial (S) and total (T) abdominal regions, TBW Total body water volume, TMM Total muscle mass, TST Triceps skinfold thickness, VAT Visceral adipose tissue, X_C Reactance

Data are expressed as mean (95% confidence interval). Different superscript letters mark statistically significant differences between groups (p < 0.05); identical letters indicate no significant difference, based on post hoc ANOVA tests

In contrast, BMI showed a modest progressive increase across tertiles, from 13.5 kg/m² in the low PhA group to 15.0 kg/m² in the high PhA group, and the difference between the extreme tertiles reached statistical significance. Similarly, AC showed a stepwise increase (16.7 cm vs. 19.0 cm in low vs. high PhA), which may be compatible with greater peripheral tissue reserves in patients with higher PhA values.

A comparable pattern was observed for TST, which nearly doubled from the low to the high PhA tertile (2.8 mm vs. 5.4 mm), suggesting greater subcutaneous fat thickness. In contrast, calf circumference did not differ significantly across groups, indicating that not all anthropometric measures showed similar descriptive patterns across PhA tertiles.

From a functional standpoint, maximal handgrip strength was significantly higher in the high PhA group compared with both low and mid tertiles, whereas the low and mid tertiles did not differ from each other. This supports an exploratory association between higher PhA and better muscle performance, without implying an independent or causal relationship.

Bioelectrical impedance vector analysis

Descriptive and progressive differences were observed across PhA tertiles in bioelectrical parameters (Table 1). As expected, PhA increased stepwise from the low to the high tertile (4.0°, 4.7°, and 5.5°, respectively), with significant differences between all groups. While R did not differ significantly between tertiles, XC showed a marked progressive increase from the low to the high PhA group (49.2 Ω vs. 73.2 Ω), with significant differences across all comparisons. Given that PhA is mathematically derived from reactance and resistance, and that resistance did not differ across tertiles, these findings indicate that the observed differences were primarily driven by variation in reactance. Because these parameters belong to the same bioelectrical framework, their interpretation should not be considered independent. As reactance is related to the capacitive properties of tissues, this may reflect differences in cellular integrity and distribution, although this interpretation should be considered indirect.

Consistently, the R-XC graph (Figure S1, Supplementary Material) was visually consistent with these findings, showing a progressive upward displacement of vectors from the low to the high PhA tertile along the reactance axis, while maintaining a relatively similar position along the resistance axis. This graphical pattern is compatible with group differences primarily driven by reactance rather than resistance and illustrates the presence of apparent bioelectrical profiles across tertiles. Notably, patients in the low PhA group were predominantly located in the lower-right region of the graph, whereas those in the high PhA tertile clustered in a higher position, a pattern that may be compatible with more favorable cellular characteristics, although this should be interpreted cautiously.

Regarding body composition, body cell mass (BCM) increased significantly across tertiles (13.6 kg, 15.8 kg, and 18.0 kg, respectively), with differences between all groups. A similar progressive pattern was observed for basal metabolism (BM) and prescribed nutritional intake, both significantly higher in the high PhA tertile compared with the lower tertiles.

By contrast, fat mass (FM), fat-free mass (FFM), total muscle mass (TMM), appendicular skeletal muscle mass (ASMM), and skeletal muscle index (SMI) did not differ significantly between groups. Likewise, overall hydration percentage and extracellular water proportion (ECW) remained comparable across tertiles.

In contrast, intracellular water (ICW) showed a significant stepwise increase from the low to the high PhA group (44.1% vs. 53.3%), while total body water percentage (TBW) was lower in the high tertile compared with the low tertile. In line with this, the upward shift observed in the BIVA vectors is compatible with a relative increase in intracellular water and potentially more favorable cellular characteristics in higher PhA groups, whereas the lower positioning of vectors in the low PhA tertile may reflect a less favorable fluid distribution profile. This pattern may indicate descriptive differences in water compartment distribution across PhA categories, despite similar total body mass, although these findings are partly embedded within the same bioimpedance-derived framework.

Muscle and abdominal ultrasound assessment

Muscle ultrasound parameters also showed significant differences across PhA tertiles. RF-CSA increased progressively from the low to the high PhA group (2.3 cm² vs. 3.7 cm²), with a significant difference between the extreme tertiles. A similar pattern was observed for the RF X-axis, which was significantly greater in the high PhA group compared with the low PhA group. The RF Y-axis also differed across tertiles, with higher values in the high PhA group compared with the low PhA group. These findings suggest that rectus femoris muscle architecture differed across PhA categories in these unadjusted comparisons, despite the absence of significant differences in body weight across tertiles. In contrast, L-SAT did not differ significantly between groups.

Regarding abdominal ultrasound measurements, T-SAT was significantly higher in the high PhA tertile compared with the low and mid tertiles. However, S-SAT and VAT did not show statistically significant differences across groups.

Biochemical profile

Most biochemical parameters did not differ significantly across PhA tertiles. Markers of glucose metabolism, renal function (urea and creatinine), protein status (total protein, albumin, and prealbumin), electrolytes (sodium and potassium), minerals (magnesium, phosphorus, and calcium), inflammatory markers (C-reactive protein), lipid profile (total cholesterol and triglycerides), vitamin B12 and iron levels were comparable between groups.

However, the Na/K ratio showed a significant progressive decrease across tertiles, with the highest values observed in the low PhA group and the lowest values in the high PhA group. In addition, folic acid levels were significantly higher in the high PhA tertile compared with the low and mid tertiles. Interestingly, TSH differed across groups, with significantly higher levels in the high PhA tertile compared with the lower tertiles. Given the number of biochemical variables assessed and the small subgroup sizes, these findings should be interpreted as unstable exploratory signals rather than robust biochemical patterns.

Clinical outcomes: length of stay and costs

Length of stay and DRG-based cost estimates showed an unadjusted inverse association with PhA at admission (Table 1; Fig. 1). Patients in the high PhA tertile experienced a significantly shorter mean length of stay compared with those in the low tertile (41.3 days; 95% CI: 31.7–51.0 vs. 58.4 days; 95% CI: 45.4–71.4; p < 0.05), representing a difference of more than two weeks between extreme groups.

Fig. 1.

Kaplan–Meier curves describing hospital stay across exploratory PhA tertiles in patients with anorexia nervosa

Consistently, estimated hospitalization costs were lower in the high PhA group (€25,829.0; 95% CI: 19,825.2–31,895.4) compared with the low PhA group (€36,523.4; 95% CI: 28,393.2–44,653.5), calculated using the standardized daily cost of €625 based on the DRG “Trastorno de conducta alimentaria.” Because these estimates were derived from daily hospitalization costs, the observed cost differences should be interpreted descriptively and are largely related to differences in length of stay. This corresponds to an approximate descriptive difference of €10,000 per admission between extreme tertiles.

Kaplan–Meier analysis (Fig. 1) showed an unadjusted descriptive pattern of shorter length of stay across increasing PhA tertiles. The low PhA group showed a slower discharge pattern, whereas the high PhA group exhibited a steeper decline in the hospitalization curve; however, this unadjusted pattern should not be interpreted as evidence that PhA or morphofunctional status explains discharge timing. The Log-Rank test showed a statistically significant difference across the overall distribution of hospitalization time (p = 0.043), suggesting differences in discharge patterns according to PhA category, although these findings should be interpreted cautiously given the limited sample size and the absence of adjustment for psychological, behavioral, and psychopathological factors that may influence hospitalization duration. The Breslow and Tarone–Ware tests showed borderline significance (p = 0.079 and p = 0.062, respectively), with consistent directional trends, which should be regarded as descriptive and exploratory patterns rather than as confirmatory evidence of shorter hospitalization among patients with higher PhA values.

Exploratory multivariate description of morphofunctional assessment in AN patients

Exploratory multivariate analyses suggested apparent descriptive differences across PhA tertiles. Given the exploratory nature of this analysis and the modest sample size, PLS-DA was used as a hypothesis-generating tool rather than for confirmatory inference. PLS-DA showed a partial separation of patients classified as low, mid, and high PhA along the first two latent components, which explained 25.6% and 22.4% of the total variance, respectively (Fig. 2A). However, this apparent separation may partly reflect mathematical or methodological interdependence among BIA-derived variables rather than independent physiological constructs. Although some overlap between groups was observed, patients in the high and low tertiles tended to cluster apart, which is compatible with descriptive differences in overall nutritional and body composition profiles according to PhA category. Consistent with the univariate results, VIP analysis identified BCMI, BCM, ECW, BMI, and body weight as the strongest contributors to descriptive group separation (Fig. 2B). However, several of these variables are directly or indirectly derived from the same bioimpedance framework, limiting their interpretation as independent contributors. In contrast, non-BIA-derived markers offered complementary context, with muscle ultrasound parameters (RF-CSA and RF Y-axis), subcutaneous fat measures (T-SAT and S-SAT), and particularly handgrip strength showing consistent differences across tertiles. These findings suggest that PhA tertiles were accompanied by measurable differences in muscle structure and function, although they do not establish an independent contribution of PhA beyond related morphofunctional variables.

Fig. 2.

A Partial Least Squares Discriminant Analysis scores, B VIP plot of nutritional and body composition parameters by PhA tertiles, and C Correlation heatmap of nutritional, body composition and ultrasound parameters with PhA subgroups

Correlation heatmap analysis (Fig. 2C) further described the pattern of associations between PhA and morphofunctional variables. Positive correlations were observed between PhA and BCMI, BCM, BMI, FM, and rectus femoris cross-sectional area, while negative correlations were observed with ECW, TBW, and hydration-related variables. Notably, the association with ultrasound-derived muscle parameters is compatible with the interpretation that PhA may be related to structural and functional tissue characteristics, rather than reflecting only electrical properties. Overall, these findings suggest that PhA stratification may reflect combined differences in cellular mass, body composition, and fluid-related parameters, rather than the effect of a single isolated variable; however, the exploratory nature of the analysis and the sample size limit definitive interpretation.

ROC analysis was performed to explore the limited and hypothesis-generating discriminatory performance of baseline PhA in relation to shorter and prolonged hospitalization (Figures S2 and S3, Supplementary Material). When prolonged stay was defined as length of stay above the cohort median (45 days), the area under the curve (AUC) was 0.649, indicating modest discriminative performance. The optimal threshold identified by Youden’s index was 4.6°, yielding 83.3% sensitivity and 60.0% specificity (Figure S2, Supplementary Material). At this cut-off, the positive predictive value (PPV) was 65.2% and the negative predictive value (NPV) 80.0%, suggesting that higher PhA values were associated with a lower probability of extended admission; however, given the small sample size and absence of external validation, these estimates should not be interpreted as clinically actionable decision thresholds. Sensitivity analysis using the 75th percentile (56 days) as an alternative definition of prolonged stay yielded comparable results (AUC 0.653) (Figure S3, Supplementary Material). A PhA threshold of 4.4° provided 82.8% sensitivity and 55.6% specificity, while maintaining the 4.6° threshold increased specificity (66.7%) with reduced sensitivity (68.9%). Across definitions, PPV ranged between 85 and 87%. Overall, these exploratory ROC analyses suggest that baseline PhA showed modest discriminatory performance in relation to hospitalization duration. Values below approximately 4.5–4.6° were associated with a higher likelihood of prolonged admission in this cohort, but these thresholds should be considered hypothesis-generating, not clinically actionable, and should not be interpreted as validated prognostic cut-offs without external validation in larger independent samples. Accordingly, PhA may be considered a descriptive complementary variable within a broader clinical and morphofunctional assessment, rather than as a standalone prognostic tool.

Discussion

The present study suggests that baseline PhA is descriptively associated with morphofunctional status and inpatient outcomes in patients with severe anorexia nervosa. However, given the observational design, modest sample size, and predominantly unadjusted analyses, these findings should be interpreted as exploratory rather than confirmatory. Importantly, these associations were observed even in the absence of significant differences in body weight across PhA tertiles, indicating that PhA may be related to morphofunctional characteristics not fully reflected by total body mass alone. Patients with lower PhA exhibited more pronounced alterations in bioimpedance parameters, reduced body cell mass, smaller rectus femoris cross-sectional area, and lower muscle strength, whereas those in the highest tertile showed more favorable structural and functional profiles. Furthermore, higher PhA at admission was associated with shorter hospitalization and lower estimated inpatient costs, although these cost differences should be interpreted descriptively because they were largely driven by length of stay; this association should not be interpreted as evidence that morphofunctional status independently explains hospitalization duration.

The close correlation between PhA and conventional nutritional indices in the present cohort is consistent with previous literature. Patients in the highest PhA tertile presented modestly higher BMI values than those in the lowest tertile; however, notably, total body weight did not differ significantly across groups. This distinction is methodologically relevant, as it indicates that PhA stratification was not exclusively explained by differences in overall mass but was also associated with descriptive differences in morphofunctional status. These findings align with Popiołek et al., who reported significant positive correlations between BMI and PhA in malnourished AN patients early in refeeding, while also demonstrating that bioimpedance parameters change dynamically and may reveal improvements in body composition not fully captured by BMI alone [51].

Beyond anthropometry, the present results suggest that PhA may reflect qualitative aspects of nutritional status. Higher PhA values were associated with significantly greater BCMI, BCM, and more favorable ICW distribution. In parallel, we observed lower ECW proportions in patients with higher PhA. Alterations in ECW and fluid redistribution are well-described features of severe malnutrition in anorexia nervosa, reflecting compromised cellular mass and membrane function [52–54]. Thus, the observed inverse relationship between PhA and ECW is consistent with the hypothesis that PhA may be sensitive to hydration shifts typical of advanced disease severity, although this interpretation remains indirect [53]. This hydration component is particularly relevant, as weight restoration alone may obscure persistent extracellular fluid expansion and incomplete cellular recovery.

Marra et al. reported that PhA is a significant predictor of resting energy expenditure in female AN patients, highlighting that higher PhA corresponds to greater metabolically active lean tissue [32]. Subsequently, Marra and colleagues showed that PhA effectively discriminates pathological leanness from constitutional leanness: anorectic patients exhibited significantly lower PhA than constitutionally lean individuals or ballet dancers despite comparable BMI [33]. Together with the present findings, where similar body weight coexisted with distinct BCM and hydration profiles, this supports the interpretation that PhA may reflect qualitative alterations not captured by quantitative weight assessment alone.

The present data also suggest that PhA may be descriptively associated with nutritional and metabolic alterations beyond single anthropometric measures. In this cohort, lower PhA values were associated with a less favorable biochemical and hydration profile, consistent with previous work. Speranza et al. identified PhA as the most reliable predictor of metabolic alterations in AN outpatients, outperforming BMI and amenorrhea duration; notably, a PhA below 4.5° was associated with significantly increased laboratory abnormalities [55]. The threshold suggested in the present study for prolonged hospitalization (4.4–4.6°) is in line with this previously proposed metabolic risk range, although the present cut-offs should be regarded as exploratory, not clinically actionable, and not as validated clinical thresholds. Likewise, De Filippo et al. reported that lower PhA (BMI < 14.5 kg/m² with 44% of the population presenting PhA < 5°) correlated strongly with anemia and neutropenia in a large AN sample, further supporting its association with malnutrition severity [56]. Taken together, these findings support the exploratory relevance of PhA as a marker associated with disease severity, while suggesting its potential role within a comprehensive assessment framework.

From a structural and functional perspective, higher PhA values were associated with greater rectus femoris cross-sectional area and higher handgrip strength, indicating more favorable muscle architecture and performance in unadjusted comparisons. These findings are consistent with Romero-Márquez et al., who reported strong correlations between rectus femoris muscle thickness and fat-free mass (r = 0.88) as well as appendicular skeletal muscle mass (r = 0.97), and demonstrated that handgrip strength improved substantially with refeeding and lean mass restoration [17]. The concordance between PhA, ultrasound muscle parameters, and functional strength in the present study is consistent with the observed PhA-related patterns, including both structural and functional components, but should not be interpreted as validation of independent clinical utility. Given that PhA is influenced by reactance, a parameter associated with membrane capacitance, it may reflect intracellular volume and aspects of cellular integrity, thereby linking hydration, cellular mass, and mechanical performance [57].

These findings also resonate with Lackner et al., who demonstrated that patients with identical low BMIs can present markedly different body composition profiles, including cases of preserved fat mass but profound muscle depletion in AN patients [8]. In such contexts, PhA may act as a complementary indicator of underlying muscle depletion not evident from weight alone.

PhA has been consistently shown to improve during recovery. Dozio et al. reported a mean increase of + 0.47° in PhA after six months of nutritional rehabilitation, paralleling gains in weight and body cell mass [58]. This dynamic responsiveness is consistent with the present observations, where higher baseline PhA was associated with more favorable structural and clinical outcomes. Nevertheless, it is important to distinguish biological recovery from psychological remission. Fortunato et al. demonstrated a disconnect between improvements in BMI or PhA and changes in eating-disorder psychopathology, indicating that somatic and psychological recovery may follow different trajectories [59]. The present findings are in line with this perspective, as they relate primarily to biological and functional domains, particularly regarding stabilization speed and resource utilization, rather than psychological outcomes. Accordingly, the absence of psychological and behavioral variables, including therapeutic engagement, personality traits, and comorbid psychopathology, limits the interpretation of hospitalization duration and recovery trajectories. Length of stay in severe anorexia nervosa should be understood as a multidimensional outcome influenced not only by physiological reserve, but also by psychological complexity, illness insight, treatment adherence, family context, and unit-specific discharge criteria [60]. Therefore, the observed association between lower PhA and longer hospitalization may partly reflect unmeasured psychological and behavioral factors, rather than poorer morphofunctional status alone.

Finally, this study observed an unadjusted association between baseline PhA and hospitalization duration. Patients in the highest PhA tertile experienced a median hospital stay more than two weeks shorter than those in the lowest tertile and incurred approximately €10,000 lower hospitalization costs. These economic differences should be interpreted descriptively, as the cost estimates were based on standardized daily hospitalization costs and therefore closely reflect differences in length of stay. To our knowledge, few studies have explored the relationship between PhA and healthcare utilization in anorexia nervosa; therefore, the present findings should be viewed as preliminary and descriptive rather than as evidence of health-economic utility. These findings align with Tarchi et al., who highlighted the need for markers capable of anticipating heterogeneity in treatment response and showed that body composition metrics, including PhA and BCM, were more informative than BMI alone in describing recovery-related patterns [34]. Taken together, the present study supports further investigation of PhA as part of a multimodal inpatient assessment framework. By capturing descriptive differences in cellular mass, hydration distribution, muscle architecture, and functional capacity despite similar total body weight in this cohort, PhA may provide complementary descriptive information, although its independent clinical relevance remains to be established. While it should not replace comprehensive clinical evaluation, it may serve as a complementary descriptive indicator of physiological status, thereby informing hypotheses for future monitoring studies in severe anorexia nervosa.

Strengths and limitations

This study is a prospective evaluation of physical recovery in severe anorexia nervosa using a multimodal and integrated morphofunctional framework, combining bioelectrical impedance vector analysis (BIVA), muscle and abdominal ultrasound, and handgrip strength. This comprehensive approach enabled the assessment of both quantitative and qualitative dimensions of nutritional status, moving beyond traditional anthropometric and biochemical markers. Importantly, PhA was associated with descriptive differences in morphofunctional profiles despite the absence of significant differences in total body weight, suggesting that it may provide complementary information to BMI-based assessment, without establishing independent clinical utility.

A major strength of this study is that PhA was consistently associated with multiple domains of nutritional status, including body cell mass, muscle architecture, and hydration-related parameters, which are characteristic of severe malnutrition in anorexia nervosa. This multimodal consistency supports its exploratory relevance as an integrative variable within a broader assessment framework, although its physiological interpretation should be considered with caution as an indirect measure and should not be interpreted as evidence of independent clinical utility. In addition, the study provides descriptive insight by linking PhA with healthcare utilization, showing that higher baseline PhA was associated with shorter hospital stays and lower estimated inpatient costs based on DRG-derived expenditure. This translational aspect should be interpreted cautiously and remains preliminary, particularly because cost outcomes largely reflect hospitalization duration.

However, several limitations should be considered. The modest sample size may limit statistical power, particularly for multivariate and ROC analyses, and increases the risk of false-positive findings given the number of variables and comparisons evaluated. This limitation is particularly relevant because several statistically significant findings were derived from small tertile-based subgroups and should therefore be regarded as exploratory signals requiring replication. The single-center design and exclusive inclusion of hospitalized females may restrict generalizability. The observational design and absence of multivariable adjustment prevent causal or independent interpretation of the associations observed; therefore, PhA should be interpreted as associated with, rather than determinative of, hospitalization outcomes. ROC analyses, performed on the entire cohort, showed modest discriminative performance; given the small sample size and lack of external validation, the derived thresholds should not be considered clinically actionable, and PhA should be considered a complementary exploratory variable rather than a standalone predictor. Additionally, the economic analysis included only direct inpatient costs derived from standardized DRG values and did not capture indirect or long-term costs. Finally, although methodological aspects of BIA have been detailed in the revised manuscript, measurement-related variability remains an inherent limitation of this technique. Moreover, psychological and behavioral factors such as illness insight, therapeutic engagement, treatment adherence, personality traits, comorbid psychopathology, family context, and discharge-related clinical criteria were not specifically assessed. Because these factors may influence length of stay and inpatient trajectory in anorexia nervosa, the observed association between PhA and hospitalization outcomes should be interpreted within this broader biopsychosocial context, rather than as a purely physiological relationship or as evidence that morphofunctional status explains hospitalization trajectory in isolation.

Despite these limitations, this study suggests that morphofunctional assessment may provide complementary descriptive information in inpatient care for anorexia nervosa and that phase angle may help characterize physiological status when interpreted within a comprehensive clinical, psychological, and organizational evaluation framework.

Conclusions

Phase angle was descriptively associated with morphofunctional status and inpatient trajectory in severe anorexia nervosa. Beyond its relationship with body cell mass, hydration distribution, muscle architecture, and functional strength, PhA was associated with length of stay and hospitalization costs in unadjusted analyses (Fig. 3). Importantly, patients with similar total body weight exhibited distinct PhA values and corresponding differences in muscle parameters, fluid balance, and discharge timing. This reinforces the limitation of relying solely on weight-based measures and supports further evaluation of cellular and hydration-sensitive parameters within routine assessment frameworks. The association between higher PhA at admission and lower DRG-based cost estimates should be interpreted cautiously, as these estimates mainly reflected shorter hospitalization duration rather than an independent economic effect. Nevertheless, this finding suggests that PhA may be related to hospitalization burden within a broader clinical and resource-use context, without demonstrating that PhA independently explains length of stay or costs. Lower PhA values may help describe reduced physiological reserve and potentially longer inpatient needs within a comprehensive clinical assessment but should not be used in isolation to guide care planning or discharge decisions.

Fig. 3.

Descriptive associations of phase angle with morphofunctional status and hospitalization-related outcomes in severe anorexia nervosa

Future multicenter prospective studies are needed to validate these findings, formally test whether phase angle provides incremental information beyond conventional anthropometric, clinical, psychological, and organizational factors, refine clinically meaningful thresholds, and evaluate its role in long-term outcomes and relapse risk. If confirmed, PhA could represent a practical adjunct to routine assessment, but its use for individualized monitoring or resource allocation requires validation in larger, externally validated cohorts.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary Material 1. ^{(2.9MB, docx)}

Author contributions

Author Contributions: Conceptualization, R.N.G.-P., A.M.-G., and J.M.R.-M.; methodology, R.N.G.-P., V.S.-G., and A.E.-E.; formal analysis, R.N.G.-P. and J.M.R.-M.; investigation, R.N.G.-P., V.S.-G., J.M.G.-B., M.N.-R., I.H.-M., and C.N.-R.; resources, A.M.-G. and M.L.-d.-l.-T.-C.; data curation, R.N.G.-P. and V.S.-G.; writing—original draft preparation, R.N.G.-P.; writing—review and editing, A.M.-G. and J.M.R.-M.; visualization, R.N.G.-P. and A.E.-E.; supervision, A.M.-G. and J.M.R.-M.; project administration, A.M.-G. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

The study was conducted in accordance with the principles of the Declaration of Helsinki. Given its observational design and minimal risk, and as all procedures were part of routine clinical practice, oral informed consent was considered appropriate and was obtained from all participants prior to inclusion. This approach was reviewed and approved by the Provincial Research Ethics Committee of Granada (approval number SICEIA-2024-003069).

Competing interests

The authors declare no competing interests.