Phase Angle as an Indicator of Depression in Maintenance Hemodialysis Patients

Patients undergoing maintenance hemodialysis (MHD) experience a variety of psychological complications. Depression is usually the most common and debilitating problem, with reported prevalence ranging from 13.1 to 76.3% [1‒3], significantly affecting quality of life, complications, treatment outcomes, and prognosis [3]. However, depression may still go undiagnosed, underdiagnosed, or untreated [4] and there is a lack of objective quantitative indicators in clinical settings. The interplay between symptoms, comorbidities, and psychosocial factors highlights the complexity and importance of managing psychosomatic disorders in the context of hemodialysis (HD) management [5]. Therefore, careful identification of depressive symptoms, early assessment of risk factors, and timely intervention are essential.

Recent researches have demonstrated that patients with depression have poorer nutritional status [6], lower physical activity, and a higher risk of sarcopenia [7], and that relationships are bidirectional. Several recent studies [8‒10] have initiated investigations into the potential associations between body composition indicators and depressive symptoms, as reliable correlations have been found between body composition and nutritional status, physical function, and muscle mass, shedding light on new avenues for understanding and monitoring depression in HD patients.

Phase angle (PhA), derived from bioelectrical impedance analysis (BIA), has emerged as a marker reflecting cellular health and integrity [11]. As a tool for assessing body composition, a number of studies have shown the strong association between PhA and nutritional status [12, 13], fluid overload [14], sarcopenia [15], and quality of life [16] in individuals on renal replacement therapy. Thus, PhA may serve as a valuable indicator not only of nutritional status, but also of broader health outcomes. Recent studies have suggested that PhA may also be related to mental health, and lower PhA has been demonstrated to be associated with an increased risk of frailty [13] and cognitive dysfunction [17]. Nevertheless, the relationship between PhA and depression remains unexplored.

In this cross-sectional study, our objective was to investigate the association between PhA and depression. Furthermore, we aim to enhance our comprehension of the complex factors influencing depression in HD patients and, hopefully, quantifying the diagnosis of depression. Investigating this association may provide insights into the interactions between depression and other complications in HD patients and inform assessment tools to improve mental health outcomes in this vulnerable population.

This multicenter, cross-sectional study enrolled MHD patients from seven HD centers in Shanghai between July 2020 and April 2021. Patients had to be over 18 years old and receive HD 4 h each time, 2 or 3 times per week, for at least 3 months (ChiCTR1900027039). Patients who were unable to complete the questionnaire or had contraindications to BIA were excluded.

This work was approved by the Ethics Committee of Tongji Hospital (K-2020-024), and the methods were carried out in accordance with the principles of the Declaration of Helsinki. All the participants were informed and signed informed consent prior to enrollment in the study.

Depressive symptoms were assessed using the Patient Health Questionnaire (PHQ-9), a validated and widely used tool for screening depression in dialysis patients [3]. The PHQ-9 consists of nine self-report questions, as follows: (1) anhedonia, (2) depressed mood, (3) sleep disturbance, (4) fatigue, (5) appetite changes, (6) low self-esteem, (7) concentration problems, (8) psychomotor disturbances, and (9) suicidal ideation, corresponding to the nine criteria for defining depression according to the Diagnostic and Statistical Manual Fourth Edition (DSM-IV) [18]. Using a 4-point severity scale (0 = not at all, 1 = a few days, 2 = more than half the days, 3 = nearly every day), the PHQ-9 measured the frequency of depressive thoughts or feelings over the past 2 weeks and summed to give a total score of 0–27 for each patient. A cut-off score of ≥10 has been shown in previous studies to be reliable and valid for defining depressive disorder in dialysis patients [19, 20]. In addition to the PHQ-9 total score, we also calculated PHQ-9 cognitive-affective subscale scores by summing items 1, 2, 6, and somatic subscale scores by summing the items 3, 4, 5, and 8 [21] to further examine the association between specific depressive symptom clusters and PhA.

Body composition was evaluated by BIA (InBody S10; BioSpace, Seoul, Korea). BIA was performed before a dialysis session. The patients removed all metal accessories, laid in a supine position, left the trunk with both arms, slightly separated the thighs, and relaxed the whole body for the measurement.

BIA measures body reactance (Xc) and resistance (R) to low-intensity electrical unit ranging from 1 to 1,000 kHz in five body segments – right arm, left arm, trunk, right leg, and left leg. BIA-derived body components, including extracellular water, total body water, fat-free mass, body fat mass, skeletal muscle index, and PhA values, were recorded. PhA of the whole body at the frequency of 50 kHz was considered [22].

Blood samples for laboratory measurements, anthropometric data, and body composition were taken on the same day before a dialysis session. Data on duration of dialysis and comorbidities were collected by self-report and review of medical records. Comorbidities were assessed using the Charlson comorbidity index (CCI) [23] and nutritional status using the malnutrition inflammation score (MIS), with a score of ≥6 indicating malnutrition [24].

Data were expressed as median with interquartile range (IQR) or numbers or percentages as appropriate. Comparisons between two groups were assessed with the nonparametric Wilcoxon test for continuous variables and the χ² test for nominal variables. Comparisons between three or more groups were assessed with the nonparametric Kruskal-Wallis test for continuous variables and χ² test for nominal variables. For the main analysis, parameters were compared according to quartiles of PhA.

Multivariable logistic regression models were mainly used to examine the association between PhA and depression. Collinearity between covariates was checked before constructing the models. Model 1 was the unadjusted model. Model 2 was adjusted for age, sex, education, single-pool Kt/V, and dialysis vintage. Model 3 added comorbidities (CCI) and nutritional status (MIS) into model 2 as adjustments. Model 4 added laboratory parameters, such as hemoglobin and serum albumin level. p value for trend was calculated by treating quartiles of PhA as a continuous variable in each model to explore the possibility of a nonlinear relationship.

We also performed restricted cubic spline to examine the association of PhA as a continuous variable with depression. The spline curve was expressed as odds ratio, adjusted for age, sex, education, single-pool Kt/V, dialysis vintage, CCI, MIS, albumin, and hemoglobin. The grouping variables for subgroup analyses included sex, the presence of diabetes mellitus (DM), education level, and nutritional status.

Receiver operating characteristic (ROC) curve and area under the curve (AUC) were used to evaluate the performance of logistic regression model and identify an optimal PhA cut-off point for the detection of depressive symptoms. The point was determined by the maximal value of Youden’s index, calculated as follows: sensitivity + specificity − 1.

All tests were two-sided, and statistical significance was set at the level of p < 0.05. All statistical analyses were performed using Stata 16.1 (Stata Corporation, College Station, TX, USA).

A total of 843 patients (median age 63 years, 61.4% male) were included in the final analysis (online suppl. Fig. S1; for all online suppl. material, see https://doi.org/10.1159/000540683). Characteristics of all patients stratified by depressive symptoms are presented in Table 1. One hundred twenty-eight patients (15.2%) had depressive symptoms, with a median age of 66 years old, 62.8% were male, 89 patients (70.1%) had cardiovascular disease (CVD), and 58 patients (46%) had DM. Compared to nondepressed patients, depressed patients were older, had more comorbidities (DM, CVD, and higher CCI score), lower educational level, and lower serum albumin level. Gender comparisons of anthropometric parameters and body composition between two groups showed that the depressed group had lower calf circumference, PhA level, higher extracellular water/total body water ratio in both sexes, lower skeletal muscle index, and fat-free mass in males.

The median PhA level (IQR) of depressed patients was 4.4° (3.9–4.9°) for males and 3.9° (3.2–4.7°) for females. The median PhA level (IQR) of nondepressed patients was 4.8° (4.2–5.4°) for males and 4.3° (3.8–5.0°) for females (Table 1).

Characteristics of patients according to PhA quartiles are shown in Table 2. With increasing PhA in each group, age, MIS, and the proportions of CVD and DM gradually decreased, whereas BMI, hemoglobin, serum albumin, mid-arm, and calf circumferences gradually increased.

The PHQ-9 total scores were 4.5 (3–9), 4 (1–7), 3 (1–7), 3 (1–6), and the prevalence of depression was 23.8%, 16.1%, 12.8%, and 8.1% in the first to fourth quartile groups, respectively. The prevalence of depression decreased significantly with quartiles of PhA increased. Correspondingly, patients in the lower PhA group had both higher cognitive-affective and somatic scores (p < 0.001, Fig. 1; online suppl. Table S1).

The unadjusted odds ratios for the logistic regression model were 3.57 (95% confidence interval [CI]: 1.98 to 6.43), 2.19 (95% CI: 1.18 to 4.06), and 1.67 (95% CI: 0.88 to 3.17) for the first to third quartiles, respectively, compared with the fourth quartile (model 1 in Table 3). The results were similar after adjustment for age, sex, education level, dialysis parameters, comorbidities, and nutritional status. The final model with adjustment of laboratory parameters revealed that the lowest quartile of PhA had a 2.19-fold higher risk of depression than the highest quartile (95% CI: 1.07 to 4.48; model 4 in Table 3). The distinguished performance of the final adjusted model for depressive symptoms was assessed by the ROC curves (Fig. 2). The AUCs for both sexes (Fig. 2a), males, and females (Fig. 2b) were 0.702, 0.686, and 0.724, respectively.

The ROC curves were plotted to identify the optimal PhA cut-off for detecting depressive symptoms. The optimal PhA cut-off was 4.9° with an AUC of 0.631 (sensitivity = 80%, specificity = 43.2%) for males and 3.5° with an AUC of 0.654 (sensitivity = 43.8%, specificity = 84.8%) for females, as shown in Table 4. Restricted cubic spline regression analysis clearly showed a relatively inverse linear relationship between PhA and depression (Fig. 3).

To evaluate the modification effects of subgroups on the association between PhA and depression, subgroup analyses of multivariable logistic regression (PhA as a continuous independent variable) were performed stratified by age (<65 or ≥65 years old), sex (male or female), diabetes (with or without), education level (<high school or ≥high school), and nutritional status (MIS <6 or MIS ≥6). p values for interactions were greater than 0.05 for all the subgroups, indicating that the increased risk of depression associated with low PhA was evident regardless of these factors (Fig. 4).

In this study, we observed that PhA levels were inversely associated with depressive symptoms in MHD patients, and this finding was not modified by age, sex, education level, nutritional status, and the presence of diabetes. Depression is the most common psychological problem in patients receiving MHD due to its relative prevalence and association with a variety of adverse clinical outcomes. In our study, the prevalence of depressive symptoms using the PHQ-9 was 15.2%. This is similar to another multicenter cross-sectional study of Chinese patients with MHD using the Beck Depression Inventory II (BDI-II), which showed a prevalence of 16.5% [9]. It is worth noting that there is variation in the prevalence of depression in the literature, partly due to differences in countries and regions, HD mode, and the assessment scales used to screen for depression [3, 20]. In addition, basic diseases, complications, medications, family support, economic status, health insurance, etc., may also contribute to the differences.

To our knowledge, this is the first large-sample, multicenter study to specifically explore the relationship between depression and PhA in MHD patients in a Chinese patient population. Our study confirmed that the lower PhA group had significantly higher PHQ-9 total scores, as well as cognitive-affective and somatic subscale scores. A previous study by Vuckovic et al. [10] showed that body composition including PhA was correlated with depression assessed by the BDI-II self-administered questionnaire in both peritoneal dialysis and HD patients. Another study by Markaki et al. [6] also found that nutritional status including PhA was associated with depression assessed by the Center for Epidemiologic Studies Depression Scale (CES-D) in HD patients. Although the sample size of these two studies was small and different depression assessment scales were used in different races, the results supported our findings and further validated the inverse association between PhA and depression.

After adjustment for confounders, PhA remained independently associated with depressive symptoms, suggesting PhA may have other properties that are also important for depressive symptoms. PhA has demonstrated the prognostic utility in multiple diseases including MHD [16, 22]. As a raw parameter of BIA, PhA measures the electrical unit of the tissue. Lower values indicate reduced cell integrity or even cell death, while higher values show an adequate health state of the whole cell membrane [25]. Pathophysiological mechanisms such as volume overload, inflammation, and oxidative stress may affect cellular health [26, 27], resulting in changes in PhA levels. These conditions make PhA a reliable and noninvasive warning indicator for HD patients.

Given that PhA is an indicator of body composition, we speculated that it might be more closely related to the somatic symptoms of depression; thus, we further analyzed the association between somatic cluster, cognitive-affective cluster, and PhA, respectively. The results showed that patients in the lower PhA group suffered both more somatic and cognitive-affective depressive symptoms. Differences in symptom improvement after effective interventions may help further clarify the strength of the link between somatic, cognitive-affective symptoms and PhA.

The cut-off values of PhA obtained from the ROC curve were ≤4.9° for males and ≤3.5° for females to detect depression in our study. So far, there is a lack of reported optimal cut-off value of PhA for depression in MHD patients. Previous studies have shown that the cut-off value of PhA for sarcopenia was 4.67° and 4.60° for males and females [28], while it was identified as a predictor of death risk at 4.50° [29]. It is important to note that PhA levels vary based on age, sex, ethnicity, and health status. Despite the lack of a universally accepted threshold for direct diagnosis of disease, the advantage of PhA lies in its real-time monitoring capability which allows for personalized assessment over time. The adjusted logistics regression model in our study showed superior discriminating performance, suggesting that in clinical practice, the combination of PhA and traditional risk factors may offer improved identification of patients at risk for depression, thereby enabling more timely and effective psychological interventions.

The study has some strengths. First, it is the largest multicenter study of PhA and depression in China to date. Moreover, we further validated the robustness of the results using restrictive cubic spline and subgroup analyses, which contribute to extrapolate our findings to similar clinical population. This study also has some limitations. This study followed a cross-sectional design. Therefore, it cannot definitively establish causal links between PhA and depressive symptoms in HD patients. Second, due to the observational nature of the study and the complexity of the disease, we were unable to rule out residual confounding, despite the inclusion of potential confounders from clinical realities and previous relevant studies.

In conclusion, our study delved into the correlation between PhA and depression, revealing an inverse association between PhA and depressive symptoms in Chinese MHD patients. We propose that PhA, as a noninvasive, reliable, and cost-effective indicator, could serve as an efficient tool for individualized assessment of depression risk in clinical practice. Subsequent research endeavors may explore the clinically significant changes in PhA through longitudinal studies and with different health interventions.

We thank all the patients who participated in this study and the medical staff who carried out the extensive clinical work and provided generous technical assistance at the multicenter dialysis unit.

The study was conducted in accordance with the principles of the Declaration of Helsinki. This study protocol was reviewed and approved by the Ethics Committee of Tongji Hospital, Approval No. K-2020-024. All the participants were informed and signed written informed consent prior to enrollment.

All authors have no conflicts of interest to declare.

This work was supported by the funding of Clinical Research Project of Tongji Hospital of Tongji University (Grant No. ITJ(QN)2204, Grant No. ITJ(QN)2106, and Grant No. ITJ(ZD)2201).

X.L., K.Z., C.Y., and Q.G. designed the present study. C.Y., W.D., J.N., J.Z., L.Z., H.Q., and S.Z. acquired the data and interpreted the results. X.L. prepared the figures and tables and drafted the manuscript. Q.G. and C.Y. revised the manuscript. All authors contributed to the manuscript and approved the final version of the manuscript.

For ethical reasons, the data generated and analyzed in this study are not publicly available. For further information, contact the corresponding author.