Clinical Research
Diagnostic Accuracy of Bioelectrical Impedance Analysis Parameters for the Evaluation of Malnutrition in Patients Receiving Hemodialysis
Nutrition in Clinical Practice Volume 00 Number 0 xxx 2018 1–12 C 2018 American Society for Parenteral and Enteral Nutrition DOI: 10.1002/ncp.10098 wileyonlinelibrary.com
Angela Teod´osio da Silva, MSc1 ; Daniela Barbieri Hauschild, MSc1 ; 1 ; Jo˜ao Luiz Dornelles Bastos, PhD2 ; Yara Maria Franco Moreno, PhD and Elisabeth Wazlawik, PhD1
Abstract Background: In the absence of a gold standard technique for assessing nutrition status in patients receiving hemodialysis (HD), we aimed to determine the diagnostic accuracy of single-frequency (50 kHz) bioelectrical impedance analysis parameters, resistance/height (R/H), reactance/height (Xc/H), and impedance/height (Z/H), and their cutoff points for malnutrition. Methods: The reference standards, Subjective Global Assessment (SGA), Malnutrition Inflammation Score, and Nutritional Risk Screening 2002, were performed at baseline and then once a year for 2 years. At least 2 assessments for each reference standard were performed during the monitoring period, and those patients who were assessed as malnourished on at least 2 consecutive occasions were classified as malnourished. Results: A total 101 patients receiving HD were evaluated. R/H and Z/H demonstrated low to moderate accuracy to diagnose malnutrition in men and low accuracy in women, whereas the accuracy of Xc/H was uncertain. The cutoff points of bioelectrical impedance vector analysis (BIVA) parameters, determined based on the SGA to maximize sensitivity and specificity simultaneously, were: R/H ࣙ330.05 and ࣙ420.92 ohms/m for men and women, respectively; Z/H ࣙ332.71 and ࣙ423.19 ohms/m for men and women, respectively. In men, sensitivity based on the cutoff points of R/H and Z/H together ranged from 73% to 89% and specificity ranged from 49% to 50%. In women, sensitivity ranged from 58% to 80% and specificity from 48% to 55%. Conclusion: BIVA parameters demonstrated low to moderate accuracy in men and low accuracy in women for the diagnosis of malnutrition. (Nutr Clin Pract. 2018;00:1–12)
Keywords bioelectrical impedance; chronic renal insufficiency; nutritional assessment; renal dialysis
Background Protein-energy wasting is characterized as the depletion of body protein and fat and is 1 of the most prevalent nutrition disorders in patients with chronic kidney disease receiving hemodialysis (HD).1 Moreover, this condition is associated with an increase in morbidity and mortality rates as well as reduction in physical function and quality of life.2,3 The nutrition status assessment and monitoring of patients receiving HD are fundamental to prevention, diagnosis, and treatment of malnutrition.1 However, a precise assessment is complex due to frequent changes in the body hydration, occurrence of comorbidities, and chronic inflammation.4,5 In the absence of a gold standard for the assessment of nutrition status, efforts have been made to identify indicators capable of an accurate diagnosis of malnutrition, such as handgrip strength (HGS), 50 kHz phase angle, and 50 kHz bioelectrical impedance vector analysis (BIVA).5-9 However, few studies have evaluated the validity of each indicator and this remains unclear.10
Bioelectrical impedance analysis (BIA) is based on the principle that body tissues offer different opposition to the passage of electrical current. Impedance (Z) refers to this opposition and is composed of 2 vectors: resistance (R) From the 1 Federal University of Santa Catarina, Health Sciences ´ Center, Post-Graduate Program in Nutrition, Trindade Florianopolis, Santa Catarina, Brazil; 2 Federal University of Santa Catarina, Health ´ Sciences Center, Public Health Department, Trindade Florianopolis, Santa Catarina, Brazil. Financial disclosures: This study was carried out with financial support from the Brazilian governmental agency CAPES (Coordination for the Improvement of Higher Education Personnel) by way of scholarships awarded to the students involved in the project. Conflicts of interest: None declared. Corresponding Author: Elisabeth Wazlawik, PhD, Federal University of Santa Catarina, Health Sciences Center, Post-Graduate Program in Nutrition, ´ Campus Universit´ario, Trindade Florianopolis, Santa Catarina, Brazil 88040–970; 55-48-3721-5138. Email:
[email protected]
2 and reactance (Xc).11 The R component of BIA represents the opposition to flow of an alternating current through ionic solutions and is inversely related to intracellular and extracellular water volume. Muscle tissue is highly conductive of electrical current due to a large amount of water and electrolytes, and presents low resistance to the passage of electric current.12 In situations of illness, a higher R value may be related to malnutrition,13 and so a higher R value may be related to the lower amount of muscle mass and, consequently, of water and electrolytes. Higher Xc values suggest a greater quantity of body cell mass and lean body mass.6,11 The parameters R, Xc, and Z were standardized by the height (H) in meters of each patient (resistance/height [R/H], reactance/height [Xc/H], and impedance/height [Z/H]) to account for the effect of the length of the conductor because the impedance of a conductor is related to its length, cross-sectional area, and frequency of the electrical current applied.14 The phase angle is calculated based on R and Xc and is considered an indicator of both nutrition status and functional state; thus, the phase angle determined at 50 kHz has been explored as a prognostic indicator in patients receiving HD.15,16 The BIVA is based on the evaluation of the values of normalized R and Xc for H. After normalization, the BIVA parameters (R/H, Xc/H, Z/H) are plotted as non-graph resistance-reactance (RXc) vectors. The position is thought to provide information about hydration status, cell mass, and cell membrane integrity.17 Some limitations of the use of BIA to evaluate nutrition status in clinical populations have been reported and have led researchers to explore the use of raw impedance data in nutrition evaluation.18-20 None of the nutrition indicators currently used are considered complete for patients receiving HD.21 Besides BIA, the International Society of Renal Nutrition and Metabolism recommends the use of anthropometric indicators, such as body mass index (BMI), percentage of fat mass (%FM), mid-upper arm circumference (MUAC) and mid-upper arm muscle circumference (MUAMC), and biochemical indicators, such as serum albumin level.1 The single-frequency (SF) BIA approach is more affordable and available than bioimpedance spectroscopy. Considering the scarcity of studies and the necessity to explore the application of BIVA parameters (R/H, Xc/H, Z/H) for the evaluation of malnutrition in patients receiving HD, this study aimed to determine the diagnostic accuracy of SF-BIA parameters in the evaluation of malnutrition in patients receiving HD using 3 reference standards: Subjective Global Assessment (SGA), Malnutrition Inflammation Score (MIS), and Nutritional Risk Screening (NRS) 2002. We also aimed to establish cutoff points for BIVA parameters, investigate correlations with other clinically important nutrition indicators (BMI, %FM, MUAC, MUAMC, HGS, and serum albumin level), and determine the prevalence of malnutrition diagnosed using BIVA.
Nutrition in Clinical Practice 00(0)
Materials and Methods Study Design and Subjects A prospective, longitudinal, diagnostic validation study was conducted with a convenience sample recruited from 2 HD clinics. The sample consisted of 101 patients aged ࣙ19 years, undergoing HD 3 times a week, and who agreed to participate from April to September 2011, June to September 2012, and July to October 2013. Exclusion criteria were patients receiving HD for 34 kg/m²; patients with amputated or atrophied limbs; those wearing a pacemaker; patients with a current diagnosis of cancer, stroke sequelae, or unable to respond; and those hospitalized for any reason and not undergoing at least 2 evaluations with the reference standards during the evaluation period. Demographic and clinical data were obtained from the medical record. In the beginning of the study, a nutrition status assessment was performed after a session of HD using different indicators: BIVA parameters (R/H, Xc/H, and Z/H), BMI, %FM, MUAC, MUAMC, and HGS. The reference standard indicators (SGA, MIS, and NRS 2002) were determined at the beginning of the study as well as once a year throughout the follow-up period. The minimum follow-up period was 10 months (1 patient) and the mean duration was 19 months (Figure 1). The clinical information, the BIVA parameters, and the reference standards were available to different evaluators, and the data were analyzed by another appraiser without influence on data collection. This study received approval by the local Research Ethics Committee under process number No. 1821 and CAAE 14375113.8.0000.0121; all participants signed a statement of informed consent.
BIA SF-BIA was performed for the determination of R and Xc R model using a portable tetrapolar device (Biodynamics 310e, Biodynamics Corporation, Seattle, WA) with a current intensity of 800 μA and at a single frequency of 50 kHz. The electrodes were placed on the midline between the prominent ends of the radius and ulna of the wrist and midline between the medial and lateral malleoli of the ankle. The patients were evaluated approximately 20 minutes following HD on the side of the body without vascular access,22 as recommended by the U.S. National Institutes of Health.23 The SF-BIA was chosen because it presents a single result for the entire body of R (ohm), Z (ohm), and Xc (ohm).
BIA parameters. R (ohm) and Xc (ohm) to obtain the Z (ohm)24 :
da Silva et al
3
Figure 1. Flowchart of sample selection process. *Body mass index >34 kg/m2 , limbs amputated or atrophied, pacemaker, cancer, heart attack, inability to respond, hospitalized or positive for acquired immunodeficiency syndrome. **Death, underwent kidney transplant, change in treatment to peritoneal dialysis, transference from dialysis center.
Z=
√
(R2 + Xc2 ).
R, Xc, and Z were standardized by H in meters: R/H, Xc/H, Z/H (ohm/m). Bioelectrical impedance vector analysis (BIVA 2002). The BIVA 2002 software developed by Piccoli and Pastori.11 was used for the determination of BIVA, with the creation of 50%, 75%, and 95% tolerance ellipses as well as confidence ellipses. The normalization of R and Xc by H was expressed as z-scores (z-R/H and z-Xc/H) with the reference population described by Piccoli et al.25 Patients with vectors outside the 75% ellipsis in the upper quadrant (underweight) and lower quadrant (cachexia) on the right side of the graph were classified as malnourished (Figure 2A). Phase angle. R and Xc for the calculation of: phase angle (◦ ) = arctangent [(Xc()/R()) × (180/π)]26
measurement parameter in the evaluation of the nutrition status. They have the advantage of being easily applicable in clinical practice and were used as a reference standard in another of our research studies.8 Patients were classified as malnourished when diagnosed with malnutrition at least 2 consecutive times by the same indicator during the followup period (2011–2013). The SGA was based on clinical history and clinical examination and classifies the patients as: A) well nourished, B) moderately malnourished or suspected of being malnourished, or C) severely malnourished. Categories B and C were grouped for the analysis.27 The MIS considers the SGA components as well as the duration of HD, comorbidities, BMI, serum albumin level, and total iron binding capacity.28 It was performed according to Kalantar-Zadeh et al, and the patients were classified as well nourished (0.90, accuracy was considered high, while 0.70–0.90 was considered moderate; 0.50–0.69 was considered low and
