The value of bioimpedance analysis in the assessment of hydration and nutritional status in children on chronic peritoneal dialysis

Abstract

Bioimpedance analysis (BIA)–body composition monitoring (BCM) has been used to evaluate the hydration and nutritional status of adults and children on dialysis. However, its clinical application still has challenges, so further exploration is valuable. We used BIA-BCM to evaluate the hydration and nutritional status of children undergoing chronic peritoneal dialysis from 1 July 2021 to 31 December 2022 in the Children’s Hospital of Fudan University to explore the clinical value of this method. A total of 84 children on chronic peritoneal dialysis (PD) were included. In the PD group, 16 (19.05%) and 31 (36.90%) had mild and severe overhydration (OH), respectively; 41.27% (26/63) had a low lean tissue index (LTI). In the PD group, patients with relative OH (Re-OH) > 5.6% had significantly higher systolic blood pressure (SBP) and SBP z score (SBPz). Patients with LTI > 12% had significantly higher body mass index (BMI) and BMI z score (BMIz). Canonical correlation analysis indicated a linear relationship (ρ = 0.708) between BIA-BCM hydration and the clinical hydration indicator and a linear relationship (ρ = 0.995) between the BIA-BCM nutritional indicator and the clinical nutritional indicator. A total of 56% of children on chronic peritoneal dialysis had OH, and 41% had a low LTI. In PD patients, SBP and SBPz were correlated with BIA-BCM Re-OH, and BMI and BMIz were correlated with BIA-BCM LTI. BIA-BCM indicators have good clinical value in evaluating hydration and nutrition.

Keywords:

• Bioimpedance analysis (BIA)–body composition monitoring (BCM)
• peritoneal dialysis (P D)
• pediatric

Introduction

Childhood chronic kidney disease (CKD) is a global public health problem with insidious onset and nonspecific and atypical clinical manifestations. Nearly 25% CKD cases may progress to end-stage kidney disease (ESKD) [1], thus requiring renal replacement therapy (RRT). Peritoneal dialysis (PD) and hemodialysis (HD) are important components of RRT. According to International Pediatric Nephrology Association data from 2018, the incidence rate of RRT among children aged 0-14 years was approximately 0.1–74.9 cases/million in the world (71 countries) and approximately 2 cases/million in China [2].

Recent studies suggest that overhydration (OH) in children on dialysis is associated with increased left ventricular mass (LVM), hypertension (HTN), and increased mortality [3]. Cardiovascular disease (CVD) is the main complication of ESKD and the leading cause of death in children on dialysis, contributing to 13.3% and 13.7% of deaths in children on PD and HD, respectively [4].

Malnutrition in children on dialysis requires the attention of clinicians because it can worsen clinical outcomes, resulting in a decreased growth rate and increased protein-energy wasting (PEW). For every 1 standard deviation decrease in height, the mortality rate increases by 14% [5]. The current diagnostic process for childhood malnutrition is complicated, and it is difficult to optimally time early interventions. In view of this, the International Society of Renal Nutrition and Metabolism (ISRNM) has emphasized the reduction in the lean tissue index (LTI) in the definition of malnutrition/PEW in recent years [6].

Bioimpedance analysis (BIA) is a simple, noninvasive body composition assessment method that obtains impedance values by measuring resistance to alternating current, denoted by R and Xc [7,8], where R is the resistance of intracellular fluid and extracellular fluid to the flow of current and is inversely proportional to tissue water content, and Xc is the capacitance generated at the interface between the cell membrane and tissue and is positively correlated with tissue cell mass [8,9]. This method is highly consistent with the gold-standard dilution method [10] and shows promise as a method to evaluate body composition in pediatrics [11]. The BIA-BCM device (Fresenius Medical Care, Bad Homburg, Germany) is a physiological tissue model based on this principle [12].

Recent studies on BIA-BCM in adults have suggested that the method can be used to assess body fluid distribution in patients on HD and can help determine dry body weight [13]. In addition, studies of adults on PD have suggested that BIA-BCM can be used to assess nutritional status [14]. Although there are some BIA-BCM studies on hydration and nutritional assessment in children on dialysis, mainly in HD [15]. There are currently not much studies on peritoneal dialysis children in China, but our study focused on Chinese peritoneal dialysis children and included a relatively large number of cases. We used BIA-BCM to evaluate the hydration and nutritional status of children undergoing chronic peritoneal dialysis in our center to explore the clinical value of this method.

Methods

Research subjects

From 1 July 2021 to 31 December 2022, 84 children received stable chronic peritoneal dialysis more than 2 months in the Department of Nephrology, Children’s Hospital of Fudan University. The study or the approval of the Research Ethics Committee of the Children’s Hospital of Fudan University [Ethics Number: 2021 (447)] and the consent of parents/caregivers were obtained.

BIA-BCM measurement

BIA-BCM was performed twice on the same day after admission (Patients with regular peritoneal dialysis are hospitalized every 3-6 months for a comprehensive assessment of peritoneal dialysis adequacy and complications of CKD in our center), and the mean value of the two measurements was taken. The same physician performed both BIA-BCM measurements using a portable BIA-BCM device (Fresenius Medical Care, Bad Homburg, Germany) [12]. Each child was placed in a semirecumbent position, and BIA-BCM electrode patches were attached to the arms. BIA-BCM measurements were performed approximately 2 h after dialysis in children on PD (Keeping empty abdomen). The output parameters included OH, relative OH (Re-OH) (namely, the ratio of OH to extracellular water (ECW)), total body water (TBW), ECW, intracellular water (ICW), ECW-to-ICW ratio (E/I), LTI, fat tissue index (FTI), etc. The upper and lower limits for normal Re-OH were defined as −7% and 7%, respectively, corresponding to the 10th and 90th percentiles for healthy individuals. Mild OH was defined as ≥ 7%, and severe OH was defined as ≥ 15% [16]. LTI was classified as normal, low, or high based on the reference range of LTI values provided by the BIA-BCM device.

Clinical indicators

The following growth and blood pressure (BP) parameters were collected for each patient: age, sex, height (H), body weight (BW), body mass index (BMI), growth rate (GR), systolic BP (SBP), diastolic BP (DBP), and pulse pressure (PP) (i.e., the difference between SBP and DBP). BW was measured by the same dialysis nurse using the same device, and the patients wore light clothing and no shoes when BW was measured. H was measured by a designated person at each center using a designated height-measuring instrument. During the study period, the measuring instrument was maintained and calibrated regularly to ensure that the readings were accurate. BW, H, and BMI were converted to z scores (BWz, Hz, BMIz), which are these values in relation to the reference values reported by the World Health Organization based on the growth curves of healthy children of the same age and sex. GR was calculated as the change in H during follow-up in the past year. To measure BP, after 5 min of resting in a quiet room, three BP measurements, at intervals of 3 min, were taken by the same pediatric nephrology nurse before the BIA-BCM measurement (for the BP readings, each child wore a cuff adjusted to a suitable size), and the mean values of the last two measurements were recorded. SBP and DBP are expressed in mmHg and z values based on the age, sex, and H of healthy children and adolescents of the same age and sex. In children, BP is associated with age, sex, and H. Normal BP was defined as SBP z score (SBPz) and DBP z score (DBPz) values less than the corresponding 90th percentiles (P90) of healthy children of the same sex, age, and H (i.e., z < 1.28), and HTN was defined as SBPz and DBPz values higher than the corresponding P90 of healthy children of the same sex, age, and H (i.e., z ≥ 1.28) [17].

Blood samples were collected within 3 days before and after the BIA-BCM assessment to measure C-reactive protein (CRP), hemoglobin (Hb), total protein (TP), pre-albumin (PAB), albumin (Alb), serum creatinine (Scr), blood urea nitrogen (BUN), triglycerides (TG), total cholesterol (TC), potassium (K), sodium (Na), and calcium (Ca).

Echocardiography was performed during routine follow-up. The left ventricular mass index (LVMI) was obtained by echocardiographers in our center using 2D m-mode echocardiography. Left ventricular hypertrophy (LVH) was defined as LVMI higher than the corresponding P90 of healthy children of the same sex and age [18].

Statistics

SPSS (version 25.0 for Windows) was used for the statistical analyses. The normally distributed measurement data are expressed as mean and standard deviation and were compared between groups by Student’s t-test. The median and interquartile spread are given for nonnormally distributed measurement data, and the Mann–Whitney U test was used for between-group comparisons. Count data are represented by n (%) and were compared between groups by the χ² test or Fisher’s exact probability test. The receiver operating characteristic (ROC) curves were plotted using SPSS according to Re-OH and LTI, respectively. According to the sensitivity and specificity of each point on the AUC curve, the point with the largest value (sensitivity + specificity-1) is calculated as the cutoff value. ROC correlation analysis was performed to identify variables with an area under the ROC curve (AUC) > 0.7, the threshold for deeming a variable correlated with the outcome. Canonical correlation analysis was used to evaluate the correlations between the clinical nutrition/hydration indicators and the nutrition/hydration indicators measured by BIA-BCM.

Results

General characteristics

This study included 84 children (mean age: 9.74 ± 4.54 years; 42 males) on chronic peritoneal dialysis between July 2021 and December 31, 2022 (Table 1). Steroid-resistant nephrotic syndrome (SRNS) (42.85%, 36/84) and congenital anomalies of the kidney and urinary tract (CAKUT) (17.86%, 15/84) were the two most common disease categories, followed by immune-related kidney diseases (10.71%, 9/84), ciliopathy (9.52%, 8/84), chronic glomerulonephritis (4.76%, 4/84), and nephrolithiasis (2.38%, 2/84). In 8 out of 84 patients (9.52%), the cause of kidney failure was unknown (Table 1). Our PD patient had a relative high proportion 47.6% (40/84) PD children with LVH. The mean duration of dialysis was 10.98 (2.38, 22.45) months. In the PD group, 26 patients (30.960%) had normal hydration (Re-OH = −7% to 7%), 16 patients (19.05%) had mild OH (Re-OH = 7% to 15%), and 31 patients (42.86%) had severe OH (Re-OH > 15%); 49.21% (31/63) of the patients had a normal LTI, 41.27% (26/63) had a low LTI, and 12.70% (8/63) had a high LTI.

Table 1. Distribution of baseline data.

	PD
N	84
Male	42 (50.00%)
Age	9.74 ± 4.54
Disease category
CAKUT	15 (17.86%)
SRNS	36 (42.86%)
Chronic glomerulonephritis	4 (4.76%)
Ciliopathy	8 (9.52%)
Nephrolithiasis	2 (2.38%)
Immune-related kidney diseases	9 (10.72%)
Other	2 (2.38%)
Unknown etiology	8 (9.52%)
Height - Z score	−1.57 ± 1.58
Weight - Z score	−1.47 ± 1.36
BMI - Z score	−1.01 (−2.20, 0.23)
SBP - Z score	2.35 ± 1.92
DBP - Z score	2.18 ± 1.79
PP (mmHg)	41.80 ± 9.51
Alb (g/L)	33.68 ± 6.76
Hb (g/L)	104.79 ± 27.70
LVMI (g/m^2.7)	43.55 (35.60, 59.80)
LVH	40 (47.60%)
Re-OH (%)	8.69 ± 17.35
Re-OH < −7% (N)	11
Re-OH −7% ∼ 7% (N)	26
Re-OH 7% ∼ 15% (N)	16
Re-OH > 15% (N)	31
LTI (kg/m^2)^a	12.55 ± 2.11
Lower LTI (N)	26
Normal LTI (N)	31
Higher LTI (N)	8

a: Due to the BIA-BCM machine detection, only 63 children with PD had valid LTI detection values. SRNS: steroid-resistant nephrotic syndrome; CAKUT: congenital anomalies of the kidney and urinary tract; PD: peritoneal dialysis; BMI: body mass index; SBP: systolic blood pressure; DBP: diastolic blood pressure; PP: pulse pressure; Alb: albumin; Hb: hemoglobin; LVMI: left ventricular mass index; LVH: left ventricular hypertrophy; Re-OH: relative OH; LTI: lean tissue index.

Correlation of Re-OH with BP

Of the 84 children on chronic dialysis, only 15.48% (13 patients on PD) showed normal BP and normal hydration (Figure 1A: Shaded portion). Seventy percent (59 patients on PD) had SBPz > 1.28 (P90), among whom 39 patients (66.20%) on PD had mild or severe OH (Quadrant I), while 20 patients (33.80%) had a normal hydration or dehydration status (Quadrant IV).

Figure 1. Correlation of Re-OH with BP. (A) Correlation of Re-OH with BP; (B; C) ROC correlation analysis of the Re-OH > 5.6% and Re-OH ≤ 5.6% subgroups in SBP value and SBPz value; (D). ROC correlation analysis. PD: peritoneal dialysis; SBP: systolic blood pressure; SBPz: systolic blood pressure z-score; Re-OH: relative OH.

The critical Re-OH value was 5.6% by ROC analysis. We divided the PD group into a Re-OH > 5.6% subgroup and a Re-OH ≤ 5.6% subgroup. The detailed clinical information and BIA-BCM indicators of the two subgroups are shown in Table S1. SBP and SBPz in the Re-OH > 5.6% subgroup were significantly higher than those in the Re-OH ≤ 5.6% subgroup (SBP: 131.10 ± 17.38 vs 115.10 ± 19.07 mmHg, p = 0.0002; SBPz: 2.82 ± 1.75 vs 1.72 ± 1.99, p = 0.0107) (Figure 1B, 1C), and the proportion of children with SBP and DBP > P90 in the Re-OH > 5.6% subgroup was significantly higher than that in the Re-OH ≤ 5.6% subgroup (p = 0.005, p = 0.03). ROC correlation analysis of the two subgroups showed that SBP [AUC 0.75 (0.64, 0.86) p < 0.0001] and SBPz [AUC 0.72 (0.60, 0.83) p = 0.001] were correlated with Re-OH, which could be used to identify the presence of OH (Figure 1D).

Correlation of LTI with BMI

Of the 63 children on PD, 47.62% (30 patients) had a normal BMIz (P5 ≤ BMIz ≤ P95) and LTI > 12 kg/m² (Figure 2A: Shaded portion), and 20.63% (13 patients) had a low BMIz (BMIz < P5) and LTI ≤ 12 kg/m².

Figure 2. Correlation of LTI with BMI. (A) Correlation of LTI with BMI. (B; C) ROC correlation analysis of the LTI > 12 kg/m² and LTI ≤ 12 kg/m² subgroups in BMI value and BMIz value; (D) ROC correlation analysis. PD: peritoneal dialysis; BMI: body mass index; BMIz: body mass index z score; LTI: lean tissue index.

The critical LTI 12 kg/m², calculated from the LTI ROC curve, was used to divide the PD group into an LTI ≤ 12 kg/m² subgroup and an LTI > 12 kg/m² subgroup (Table S2). Comparative analysis of the LTI ≤ 12 kg/m² and LTI > 12 kg/m² subgroups indicated that the LTI ≤ 12 kg/m² subgroup has lower BMI and BMIz values (BMI: 14.54 ± 2.42 vs 16.25 ± 3.33, p = 0.0298; BMIz: −1.89 ± 1.74 vs −0.80 ± 1.82, p = 0.0168) (Figure 2B, 2C). ROC correlation analysis indicated that BMI [AUC 0.75 (0.61, 0.89) p = 0.005] and BMIz [AUC 0.72 (0.57, 0.86) p = 0.001] were correlated with LTI, which can be used to identify the presence of a nutritional imbalance (Figure 2D).

Canonical correlation analysis of the relationship between the BIA-BCM hydration indicator and the clinical hydration indicator in the PD group

Taking OH, Re-OH, ECW, ICW, and E/I as the BIA-BCM hydration indicators (variables Y and Y1-Y5 in sequence) and DBP, PP, Hb, SBP, SBPz, and LVMI as the clinical hydration indicators (variables X and X1-X6 in sequence), canonical correlation analysis was used to establish a model. A total of five pairs of canonical variables were obtained (Table S3, 4). The correlation test of the five pairs of canonical variables (Table S5) indicated that only the first pair of canonical variables had a strong linear relationship (ρ = 0.708, p < 0.0001). Therefore, the first pair of canonical variables was used to construct functions. That is, the formula for the Y variable was V1 = 0.253Y1 − 0.478Y2 − 0.716Y3 − 0.321Y4 + 0.049Y5, and the formula for the X variable was U1 = −1.894 × 1 − 1.101 × 2 + 0.122 × 3 + 0.363 × 4 + 1.29 × 5 + 0.159 × 6 (where a negative coefficient indicates a negative relationship and a positive coefficient indicates a positive relationship). The contribution rates of the BIA-BCM hydration indicator and clinical hydration indicator are shown in Figure 3A.

Figure 3. Canonical correlation analysis of the relationship between the BIA-BCM indicator and the clinical indicator in the PD group. (A) BIA-BCM hydration indicator and the clinical hydration indicator. (B) BIA-BCM nutritional indicator and the clinical nutritional indicator. BIA-BCM, bioimpedance analysis-body composition monitor; PD: peritoneal dialysis; Hb: hemoglobin; LVMI: left ventricular mass index; SBPz: systolic blood pressure z-score; SBP: systolic blood pressure; DBP: diastolic blood pressure; PP: pulse pressure; OH: overhydration; Re-OH: relative OH; ECW: extracellular water; ICW: intracellular water; E/I, ECW to ICW ratio; CRP: C-reactive protein; TC: total cholesterol; TG, triglyceride; PAB, prealbumin; TP: total protein; Alb: albumin; HZ: heigh z-score; GR: growth rate; WZ, weight z-score; BMI: body mass index; BMIz: body mass index z score; LTI: lean tissue index; LTM: lean tissue mass; FTI: fat tissue index; ATM: adipose tissue mass.

Canonical correlation analysis of the relationship between the BIA-BCM nutritional indicator and the clinical nutritional indicator in the PD group

Taking LTI, FTI, lean tissue mass (LTM), FAT, and adipose tissue mass (ATM) as the BIA-BCM nutritional indicators (variables Y and Y1-Y5 in sequence) and BMIz, total protein, triglycerides, total cholesterol, Hb, CRP, Alb, PAB, GR, Hz, and BWz as the clinical nutritional indicators (variables X and X1-X11 in sequence), canonical correlation analysis was used to establish a model. A total of five pairs of canonical variables were obtained (Table S6, 7). The correlation test of the five pairs of canonical variables indicated (Table S8) that the first and second pairs of canonical variables had a strong linear relationship (ρ = 0.995, p < 0.0001) for the first pair of typical variables and could be used for typical variable construction functions. That is, the formula for the Y variable was V1 = −0.716Y1 − 0.616Y2 + 0.059Y3 − 0.06Y4 − 0.097Y55, and the formula for the X variable was U1 = −1.059 × 1 − 0.031 × 2 − 0.03 × 3 − 0.023 × 4 − 0.084 × 5 − 0.018 × 6 − 0.032 × 7 + 0.074 × 8 + 0.022 × 9 − 0.024 × 10 + 0.086 × 11 (where a negative coefficient indicates a negative relationship and a positive coefficient indicates a positive relationship). The contribution rates of the BIA-BCM nutritional indicator and clinical nutritional indicator are shown in Figure 3B.

Discussion

In our center, 56% of children on chronic PD had OH, and 41% had a low LTI. SBP and SBPz were correlated with BIA-BCM Re-OH in children on PD, and BMI and BMIz were correlated with BIA-BCM LTI in children on PD. Hence, BIA-BCM indicators have good clinical value in evaluating hydration and nutrition in children.

According to the European guidelines for pediatric dialysis, PD is the first choice for RRT in infants, and HD is often used in adolescents [19]. Patients on PD or HD are prone to acute and chronic OH [20–22]. In our PD group, according to the BIA-BCM results, 19% and 43% of the patients had OH and severe OH, respectively. These results are similar to the findings of a study by Karava et al. where 48% and 20% of children on PD had OH and severe OH, respectively.

HTN is one of the most common and most dangerous complications in patients on chronic dialysis. OH can affect BP status, including the SBP [23] and PP [24]. A correlation between BP status and Re-OH has been demonstrated in three recent studies that included children [25]. In this study, when comparing the normal hydration group with the mild OH group and the severe OH group, there was no correlation between Re-OH and the above values, though SBP and SBPz were significantly higher in the Re-OH > 5.6% subgroup than in the Re-OH ≤ 5.6% group in our study. In addition, the clinical indicators SBP and SBPz were correlated with the hydration indicator Re-OH. Nearly half of the hypertensive children in the PD and HD groups in this study did not have OH, suggesting that other causes of HTN need to be explored to choose an appropriate treatment. Similar observations were reported in a multicenter pediatric study that evaluated 463 patients undergoing predialysis treatments, where approximately one-third of the patients had elevated BP and normal or low Re-OH [26]. This finding suggests that elevated SBP is not always related to OH and that there may be volume-independent factors underlying OH [27]. Vascular stiffness, congestive heart failure, residual renal function, and underlying renal disease can contribute to differences in hydration measurements and BP [28]. Therefore, BP alone is not an effective measure of the hydration status of patients on chronic dialysis. Incorporating BIA-BCM measurement indicators into the clinical evaluation indicators can help clinicians distinguish volume-dependent and volume-independent HTN and avoid excessive ultrafiltration, which would be a useful tool combined with clinical judgment in assessing overhydration.

The incidence rate of malnutrition in patients on chronic dialysis is high. Three characteristics of malnutrition in children with CKD, namely, PEW, obesity, and growth disturbance, are among the predictors of mortality in patients on chronic dialysis and are associated with muscular atrophy, inflammation, atherosclerosis, and other serious diseases [29]. The prevalence rates of PEW in the CKD population aged 1-16 years in the United States range from 6 to 65%. The current diagnostic process for PEW is complex, making early identification of PEW difficult. In 2014, the ISRNM revised the diagnostic criteria for PEW in children based on the adult criteria. They include five parameters: reduced BMI, reduced mid-arm circumference (MAC), blood biochemistry (Alb, cholesterol, transferrin, CRP), decreased appetite, and short stature [30]. Due to the large error in MAC measurements, this indicator is not recommended by the Kidney Disease Outcomes Quality Initiative (KDOQI) guidelines. In 2013, the ISRNM defined PEW as a state of reduced storage of protein and energy fuel (protein and fat mass) in the body [31] that results in changes in adipose tissue mass and lean tissue mass. This definition highlights the early warning and diagnostic role of lean tissue mass in malnutrition in patients on dialysis. In 2020, the BIA-BCM results of a 2-year follow-up of 131 adult patients on PD showed that all-cause mortality in the low-LTI group was significantly higher than that in the normal LTI group and that all-cause mortality increased with a decrease in LTI. However, changes in lean tissue mass may go unnoticed in clinical assessments of patients on chronic dialysis. Unlike changes in BW and BMI, changes in body composition are a dynamic process, so monitoring changes in body composition, rather than a single measurement, can more accurately reflect the body composition. In this study, canonical correlation analysis indicated that there was a strong correlation between BIA-BCM nutritional indicator data and clinical nutrition indicator data, and the ROC correlation results suggested that the LTI measured by BIA-BCM was correlated with the BMI and BMIz values of children. Therefore, the combination of growth parameter measurement and BIA-BCM measurement indicators can better evaluate the adipose tissue and lean tissue of individual patients and accurately evaluate nutritional status. In this study, approximately 60% of patients on PD had a low LTI and/or a low BMI. Clinicians should be aware of this situation and provide timely nutritional interventions.

This study has certain limitations. Only the baseline data of patients on chronic dialysis receiving BIA-BCM for the first time were included in this study; no follow-up data of these indicators were included. In this study, the effectiveness of the BIA-BMC method could not be evaluated because there was no gold standard for assessing hydration, but only suggested the correlation between BIA-BCM indicators and clinical indicators. At the same time, we should include in detail the antihypertensive drugs and residual kidney function of children with hypertension. Due to the lack of reference data from Asian children, the reference interval for LTI in this paper is based on machines derived from 1000 Caucasian adults.

This relatively comprehensive study performed BIA-BCM to assess the clinical hydration and nutrition status of PD children. In the future, larger, multicenter, randomized controlled studies should be conducted to provide better evidence for or against using BIA-BCM to assess the hydration and nutrition of PD children.

We thank all participating patients and their families

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

We thank all participating patients and their families. Additionally, this work was supported by the Shanghai “Science and Technology Innovation Action Plan” (Sailing Special Project) (22YF1403600).