Heart failure (HF) is associated with neurohumoral activation and abnormalities in autonomic control that lead to sodium and water retention and increased excretion of potassium, with subsequent edema or ascites.1 As a consequence, evaluation of hydration state is a relevant component in the assessment and treatment of patients in the emergency department (ED) with acute decompensated heart failure (ADHF).2
Although clinical assessment remains the mainstay of estimating the hydration state of patients with HF, it could be challenging to evaluate subtle hyperhydration or dehydration, which may result in increased short- and long-term morbidity. Consequently, the main goal of ED physicians is to distinguish immediately among (1) normal hydrated, (2) dehydrated, and (3) hyperhydrated patients.3–6 Thus, it is important to use an easy, reproducible, and noninvasive method for the estimation of fluid compartments.7,8
The bioimpedance vector analysis (BIVA) is a noninvasive technique to estimate body mass and water composition by bioelectrical impedance measurements, resistance, and reactance,9 which are already used in clinical practice for evaluating the fluid content in dialyzed patients.10 Brain natriuretic peptide (BNP) is an important biomarker released from the ventricles in response to wall stretch by increase in either volume or pressure during the acute phase of HF. Plasma concentrations of BNP correlate with symptomatic New York Heart Association (NYHA) class as well as prognosis in patients with systolic HF.11
The aim of this study was to verify whether BIVA could be a valid methodology to assess the fluid overload, whether there was a correlation between BNP and fluid overload detected by BIVA, and whether this technique coupled with BNP assessment could be a valid guide in the management of therapy in patients with ADHF.
Materials and Methods
A total of 51 patients arriving to the ED of the University Hospital Sant’Andrea (Rome, Italy) from December 2008 to April 2009 were enrolled.
The patients were divided into 3 groups. Inclusion criteria for the first group (n=25) were acute respiratory distress (of at least 3 days), BNP level >500 pg/mL, clinical evidence of ADHF, and history of chronic HF with NYHA class 3 or 4 disease. The main characteristics were male/female, 11/4; age, 81±6 years; body mass index (BMI), 25±3 kg/m2; serum creatinine, 1.5±1 mg/dL; systolic blood pressure (SBP), 156±41 mm Hg; diastolic blood pressure (DBP), 85±26 mm Hg; heart rate (HR), 85±17 beats per minute; respiratory rate (RR), 23±7 breaths per minute; BNP, 1011.5±524.2 pg/mL; hydration, 79%±6%; and vascular pedicle width (VPW), 79±8 mm. Inclusion criteria for the second group (n=13) were acute respiratory distress (of at least 3 days), BNP level <400 pg/mL, no clinical evidence of ADHF, and every noncardiogenic dyspnea. The main characteristics were male/female, 7/6; age, 78±7 years; BMI, 26±5 kg/m2; serum creatinine, 1.5±1.3 mg/dL; SBP, 137±39 mm Hg; DBP, 70±8 mm Hg; HR, 87±23 beats per minute; RR, 26±6 breaths per minute; BNP, 235±72 pg/mL; hydration, 75%±47%; and VPW, 71±10 mm. The patients in the last group (n=13) had no clinical findings of cardiac or respiratory diseases and had BNP levels <100 pg/mL. The control group comprised this group together with group 2. The main characteristics were male/female, 6/7; age, 81±3 years; BMI, 25±2 kg/m2; serum creatinine, 1.2±0.2 mg/dL; SBP, 125±13 mm Hg; DBP, 70±5 mm Hg; HR 65±5 beats per minute; RR, 15±3 breaths per minute; BNP 20±22 pg/mL; hydration, 74%±2%; and VPW, 64±13 mm. Patients were stratified for sex, age, and BMI (Table I).
Table I. Patient Characteristics
Dyspnea and ADHF (n=25)
Abbreviations: ADHF, acute decompensated heart failure; BMI, body mass index.
The exclusion criteria for each group were age younger than 18 years, thoracic and abdominal trauma, pneumothorax, past pneumonectomy, and invasive or noninvasive ventilation; pregnant women; patients unable to give written consent; and patients with intraperitoneal fluid, Glasgow Coma Scale <8, BMI >30 kg/m2, estimated glomerular filtration rate <60 mL/min, liver failure, constrictive pericarditis or cardiac tamponade, and exitus before 72 hours.
In our study, the methodology used to analyze body fluids was BIVA, a noninvasive technique to estimate body mass and water composition by bioelectrical impedance measurements, resistance, and reactance.10 Resistance (R o Rz) is the strength opposed by a tissue to the electric current, conducted by electrolytes; fat-free tissues and fluids are good conductors, while bone tissue and fat tissues are bad conductors. Reactance (X o Xc) is due to the presence of inductors and/or electrolytic capacitors, and that is an indirect measure of cell membranes’ integrity.12,13 The arc tangent of Xc/R is called the phase angle. The patient lies supine, without metal contacts, with straddle inferior limbs at 45° and superior limbs abducted at 30° to avoid skin contacts with the trunk.12 Two skin electrodes are applied, one on the right hand and the other on the right foot. A minimal interelectrode distance of 5 cm has been recommended to prevent interaction between electrodes. The patient was laid recumbent on a nonconductive surface. Impedance measurements were taken independently from fasting as free fluid within the trunk (ascites, urine, food, or milk) does not contribute to the Xc component of impedance.
BIVA quickly evaluates body mass in normal, extremely obese, and undernourished patients and assesses the hydration state in normal (hydration, 72.7%/74.3%), hyperhydrated, and dehydrated patients by drawing a graphic (hydragram) representation of the results.14,15 BIVA shows Rz and Xc values adjusted for the patient’s height and plotted on the interpretative scheme, related to sex and age16 (Figure 1). Mean vector displacements in the RXc plane indicate a definite change in body composition of groups of patients, and 2 mean vectors with separate 95% confidence ellipses indicate a statistically significant difference in vector position on the RXc graph. Short vectors with a small phase angle are associated with edema, whereas long vectors with an increased phase angle indicate dehydration.17
BIVA measurements were performed with electrofluidgraphy.18
As guidelines recommend, we performed the chest radiography as a diagnostic procedure to every dyspneic patient.19 This technique is used worldwide to evaluate lung congestion and also for differential diagnosis between cardiogenic and noncardiogenic dyspnea. Moreover, we evaluated VPW to obtain information about the volemic status of the patient.20 VPW was measured by dropping a perpendicular line from the point at which the left subclavian artery exits the aortic arch and measuring across to the point at which the superior vena cava crosses the right mainstem bronchus.21
Another noninvasive determination of fluid content used in our study was the caval index (CI).This parameter was acquired by an ultrasonographic assessment of vena cava dimensions. It was measured while patients were in the supine position. Inspiratory inferior vena cava and expiratory inferior vena cava diameters were measured 2 to 3 cm from the right atrial border in a long-axis/subxiphoid view with a 2- to 4-MHz curvilinear probe. Measurements were taken during a normal respiratory cycle; images were frozen on the ultrasonography machine and frame-by-frame analysis was performed to determine both expiratory inferior vena cava and inspiratory inferior vena cava values. The inferior vena cava and caval index was calculated as the relative decrease in inferior vena cava diameter during a normal respiratory cycle (expiratory inferior vena cava – inspiratory inferior vena cava / expiratory vena cava) and was also expressed as the inferior vena cava and CI percentage (inferior vena cava and CI × 100%).22 The relationship was the collapsibility index.23
A blood sample was collected for BNP measurement for all patients, as the literature24 has demonstrated that it is one of the most sensitive biomarkers to predict the prognosis and to make a rapid diagnosis of ADHF.25 BNP plasma concentration measurement was performed with the Triage BNP kit (Biosite, Inc, San Diego, CA), a single-use, ready-to-use, florescence immunoassay. A venous blood sample in an ethylenediaminetetraacetic acid tube was collected and measured in the Triage BNP tester.11
Hydration status, measured by BIVA, BNP, and CI, was evaluated at admission to the ED, at 24 and 72 hours, and at discharge. VPW was assessed only at admission and discharge. A phone call follow-up at 3 months was made to all discharged patients.
All analyses were performed using MedCalc for Windows (release 10.4.8; MedCalc Software, Mariakerke, Belgium). Differences between 2 independent groups of continuous variables were compared using the Mann–Whitney U test. The correlation between BNP and BIVA was calculated using Spearman rank test. The correlation between urine output and BIVA was calculated using Spearman rank test. P values <.05 were considered significant. A receiver operating characteristic (ROC) curve assessing the ability of BIVA and VPW to predict events at 3 months was plotted.
BIVA of ADHF patients at admission to the ED was compared with BIVA of controls, and the difference was statistically significant (P<.0007); ADHF patients had greater hydration (76.7%±4.0%) compared with controls (73.1%±1.9%). The hydration state of ADHF patients was evaluated at different points during hospitalization using BIVA. Baseline BIVA mean value at admission (t 0) was 76.74±4.0, BIVA mean value at 24 hours was 76.0±4.0, BIVA mean value at 72 hours was 75.3 ± 4.0, and BIVA mean value at discharge was 74.4±2.0. The difference between t 0 and t 72 BIVA mean values (P<.001) and between t 0 and discharge BIVA mean value (P<.0001) was statistically significant. The BIVA mean value in the control group was not statistically significant (P=.9372) (Figure 2). VPW in ADHF patients (77±7 mm) showed a statistically significant difference (P<.0004) compared with VPW in controls (48±5 mm). VPW mean values in ADHF patients on admission to the ED (77±7 mm) compared with VPW mean values at discharge (65±10 mm) were statistically significant (P=.003). As demonstrated by hydration state, there were no significant differences between VPW values at admission (48±5 mm) and at discharge in the control group (46±3 mm), showing no congestive state. CI in ADHF patients (29±20) showed a statistically significant difference (P=.02) compared with CI in controls (60±8). Baseline CI mean value at admission (t 0) was 29±20, CI mean value at 24 hours was 37±25, CI mean value at 72 hours was 42±24, and CI mean value at discharge was 51±17. CI mean values in ADHF patients on admission to the ED compared with mean values at discharge were statistically significant (P<.001).
Values of BNP, BIVA, CI, and VPW at different considered times are shown in Figure 2.
Patients’ hydration state was immediately evaluated by BIA-Vector (Figure 3). The values' shift from a hyperhydrated state to a normal hydrated state showed improvement due to diuretic treatment in ADHF patients. BNP and BIVA at admission in ED were correlated in ADHF patients (Figure 4). The BIVA ROC curve showed a normal hydration state at 24 hours (Figure 5), with a specificity of 100% (area under the curve, 0.78; P=.04, 24 hours). In patients with average hydration values >80.5%, there was a correlation with events at 3 months (death or rehospitalization for cardiogenic event) with a sensitivity of 22% and specificity of 94.2% (+likelihood ratio [LR] 4.6 + predictive value [PV] 66.7 – PV 74.1). There was also a significant statistical value (r=.41; P=.01) between BNP and VPW (Table II).
Moreover, VPW >82 mm has been correlated with events at 3 months (death or rehospitalization for cardiogenic events) with a sensitivity of 54% and specificity of 84.2% (+LR 3.4 + PV 66.7 – PV 76.2) (Figure 6). Furthermore, no statistically significant correlation was found with the other evaluated parameters for outcomes: sex, age, serum creatinine, blood pressure, HR, RR, and BMI.
Currently, ADHF diagnosis is based on laboratory tests (complete blood cell count, sodium, potassium, glucose, blood urea nitrogen, serum creatinine, and arterial blood gas count and biomarkers such as troponin I and BNP),26,27 chest radiography, thoracic ultrasonography,28 respiratory variation of inferior vena cava diameter,29 echocardiography,30 and invasive hemodynamic monitoring. Most of these techniques are expensive, are not always available, and need time to be performed. In this setting, BIVA represents a valid, rapid, noninvasive, inexpensive technique in the evaluation of hydration state. In patients with hyperhydration due to ADHF, some studies have shown that reactance is strongly related to BNP values and NYHA severity class.31,32 Other studies have found a correlation between impedance and central venous pressure in critically ill patients.33 Our clinical experience with BIVA showed positive results when performed in ADHF patients admitted to the ED. Together with BIVA, we evaluated BNP34 and ultrasonographic measurement of CI and VPW obtained through chest radiography.35 Hydration state has been analyzed at admission to the ED, at 24 and 72 hours, and at discharge, and BIVA measurements have been correlated with the other 3 tests (Figure 2). Moreover, to distinguish cardiogenic dyspnea from noncardiogenic dyspnea, it is important to assess BNP values,36,37 which allow discriminating cardiogenic dyspnea for values >500 pg/mL.38 In our experience, the combined use of both BIVA and BNP improves diagnosis and therapeutic management in ADHF patients, maintaining the optimal hydration state through the correct use of diuretics (Figure 3 and Figure 4). As a matter of fact, the finding of elevated BNP levels coupled with BIVA short vector from admission to discharge leads the ED physician to a continuous and noninvasive evaluation of the fluid overload and allows appropriate management of diuretic therapy to prevent iatrogenic dehydration.
Our study also shows that VPW evaluation by chest radiography has very high specificity for values >80 mm to distinguish cardiogenic dyspnea from noncardiogenic dyspnea.
Unlike in ADHF patients, in the control group we found no statistically significant differences between hydration state values at admission to the ED and at discharge, showing the noncongestive state of patients with noncardiogenic dyspnea. These data are clearly described by the BIA-Vector (Figure 3).
In our preliminary results, BIVA and CI showed a significant and indirectly proportional correlation at admission to the ED and at 24 and 72 hours. These data confirm the strong correlation between hyperhydration and central venous congestion. The correlation between BIVA and urine output at 72 hours became directly proportional. It confirms a good response to diuretic therapy with the shift of fluids from interstitial spaces: the more patients were hyperhydrated, the more urine output was conspicuous. BIVA’s utility is confirmed by the ROC curve that shows a normal hydration state at 24 hours (Figure 5) and 72 hours, with a specificity of 100% (area under the curve, 0.78; P=.04, 24 hours; area under the curve, 0.9; P=.0001, 72 hours). The efficacy of diuretic therapy and the effectiveness of BIVA measurements were confirmed by normalization of BNP, VPW, and CI values at discharge (Figure 2). These results suggest that BIVA, more than clinical signs, could be useful to manage the diuretic therapy in ADHF patients in the ED. Therefore, there is a correlation with events at 3 months (death or rehospitalization for cardiogenic event) in patients with average hydration values >80.5% in the same way as there is in patients with VPW >82 mm.
The combined use of both BIVA and BNP improves the management of ADHF patients in the ED. It allows a faster and more accurate diagnosis, it is a valid support to distinguish cardiogenic dyspnea from noncardiogenic dyspnea, and it contributes to success in treatment. The combined use of both BIVA and BNP supports ED physicians’ decisions about diuretic therapy, preventing excessive dehydration. Moreover, BIVA was able to identify that patients with a higher percentage of hydration have more death and/or rehospitalization probabilities.
Disclosures: The authors have no financial relationships with any commercial interests. Dr. Di Somma received an honorarium, funded by an unrestricted educational grant from Abbott Laboratories and Otsuka America Pharmaceuticals for time and expertise spent in preparation of this article.