a b s t r a c t

Introduction: The aim of this study was to evaluate bioimpedance vector analysis (BIVA) for the diagnosis of Acute heart failure in patients presenting with acute dyspnea to the emergency department (ED).

Methods: Patients with acute dyspnea presenting to the ED were prospectively enrolled. Four parameters were assessed: resistance (R), reactance (Ra), total body water (TBW), and extracellular body water (EBW). Brain natriuretic peptide measures and cardiac Ultrasound studies were performed in all patients at admission. Patients were classified into AHF and non-AHF groups retrospectively by expert cardiologists.

Results: Seventy-seven patients (39 men; age, 68 +- 14 years; weight, 79.8 +- 20.6 kg) were included. Of the 4 BIVA parameters, Ra was significantly lower in the AHF compared to non-AHF group (32.7 +- 14.3 vs 45.4 +- 19.7; P b .001). Brain natriuretic peptide levels were significantly higher in the AHF group (1050.3 +- 989 vs 148.7 +- 181.1 ng/L; P b .001). Reactance levels were significantly correlated to BNP levels (r = -0.5; P b .001). Patients with different mitral valve Doppler profiles (E/e’ <= 8, E/e' >=9 and b 15, and E/e’ >= 15) had sig- nificant differences in Ra values (47.9 +- 19.9, 34.7 +- 19.4, and 31.2 +- 11.7, respectively; P = .003). Overall, the sensitivity of BIVA for AHF diagnosis with a Ra cutoff at 39 ? was 67% with a specificity of 76% and an area under the curve at 0.76. However, Ra did not significantly improve the area under the curve of BNP for the diag- nosis of AHF (P = not significant).

Conclusion: In a population of patients presenting to the ED with dyspnea, BIVA was significantly related to the AHF status but did not improve the diagnostic performance for AHF in addition to BNP alone.

(C) 2015

1. Introduction

Dyspnea is a common symptom for emergency department (ED) consultation. It is also the most common symptom at presentation for acute heart failure [1-3]. However, this symptom is not specif- ic and is present in many other non-cardiac-relatED diseases. Acute heart failure represents the most common cause of hospitalization through the Medicare system in the United States. Most of these hospi- talizations are secondary to chronic heart failure (CHF) decompensation [3,4].

Several reports show that an accurate diagnosis with early and ap- propriate therapeutic management improves the prognosis of patients with AHF [5,6]. Fonseca [7] showed that only 25% to 60% of AHFs were correctly diagnosed in the ED. In addition, symptoms and clinical pre- sentation for this pathology have limited sensitivity and specificity

* Corresponding author.

E-mail address: [email protected] (N. Genot).

[1,3]. There is a need for additional tests to improve diagnostic capacity and enhance timely therapeutic management.

Brain natriuretic peptide (BNP) testing has emerged as an efficient diagnostic tool for AHF detection [8,9]. However, its diagnostic perfor- mance can be limited in selected subgroups of AHF patients. Patients with CHF may paradoxically have high or low values of BNP [8-10]. Sim- ilarly, BNP in “gray zone” values also raise problems of interpretation. Other noninvasive HF diagnostic tools such as chest radiography or car- diac ultrasound study also have limitations in terms of either diagnostic performance or ED availability [2,11,12].

One of the proposed methods for the diagnosis of AHF and evalua- tion of body fluid overload is electrical bioimpedance [13]. It is quick, reproducible, noninvasive, and inexpensive. The impedance of a patient characterizes the opposition that body tissues present to an electric current’s transmission. Bioimpedance has been studied in various fields such as nutrition [14], nephrology [15], hepatology [16], and cardiology [17-19].

Preliminary clinical studies in selected groups of patients have shown a significant relationship between impedance parameters and AHF [20-24]. The fluid overload present in AHF leads to a decrease in

http://dx.doi.org/10.1016/j.ajem.2015.04.021

0735-6757/(C) 2015

diagnostic workup“>body impedance. Measurement of patient’s bioimpedance in the ED could therefore be relevant in the AHF diagnostic workup.

The principal aim of our study was to evaluate the diagnostic perfor- mance of bioelectrical impedance vector analysis in patients admitted to the ED for dyspnea and compare it to BNP and echocardiographic pa- rameters for the accurate diagnosis of AHF.

1. Material and methods
   1. Study population

This study was a prospective monocentric observational study. The study was conducted from February to June 2013 at the Emergency cardiology consultation (ECC) at the University Hospital of L. Pradel at Bron (France), a tertiary referral center. Our institution is specialized in cardiac and pulmonary diseases, and the attached ED is specialized in cardiac and pulmonary emergencies and symptoms. Patients who consult in this ECC are referred by themselves, their general practition- er, or the mobile intensive care units (911 services in France). All con- secutive patients presenting to the ECC with symptoms of dyspnea were eligible for inclusion.

Exclusion criteria were as follows: acute coronary syndrome, age younger than 18 years, refusal to participate in the study, and pregnancy. Patients who satisfied the inclusion criteria were informed of the study protocol. Free and informed consent was obtained from each pa- tient included in this study. The study protocol was approved by our In-

stitutional Review Board.

Diagnostic workup

For each patient, dyspnea was assessed according to the New York Heart Association classification. Previous medical history and baseline clinical and Biological characteristics were reported. Chest radiography was performed at the discretion of the attending physician at the time of admission in the ECC.

Serum BNP levels (immunoassay method; Architect I2000) were assessed upon admission. The upper threshold of normal value used for the diagnosis of AHF according to our local laboratory was greater than or equal to 100 ng/L [9].

A transthoracic Cardiac ultrasound scan (GE Vivid S5; General Elec- tric, Milwaukee, WI) was performed in all patients. Left ventricular ejec- tion fraction was measured, and left ventricular end-diastolic pressure (LVEDP) was assessed according to the E/e’ ratio obtained by Doppler tissue imaging at the mitral annulus in the lateral wall. The LVEDP clas- sification was based on the American Society of Cardiology ultrasound guidelines [25]. An E/e ‘ ratio less than or equal to 8 was classified as nor- mal LVEDP (class I), a ratio between 9 and 14 was classified as interme- diate LVEDP (class II), and a ratio greater than or equal to 15 was considered as elevated LVEDP (class III). If LVEDP was not possible to as- sess (atrial fibrillation, mitral valve disorder, severe segmental wall mo- tion), patients were arbitrarily included into the class II category.

The bioimpedance measurements were performed with a Z-Metrix Bioparh?m device (Z-Metrix, Bioparh?m, Bourget-du-Lac, France) at patient’s admission. Four electrodes were placed on the patient’s body: 2 on the right hand (between the metacarpals) and 2 on the side of the right calf (1 above the Lateral malleolus and 1 below). Elec- trodes were placed apart by a space of approximately 4.0 cm. The mea- surement was performed on arrival at the ED, before initiation of any therapy, and was completed in supine position. Each analysis took 15 seconds, and results obtained were recorded electronically. The dif- ferent parameters measured were total body water (TBW), extracellular water (EEW), reactance (Ra), and resistance (R).

To assess normal values of bioimpedance measures, we also per- formed bioimpedance measurements (TBW, ECW, Ra, and R) of 14 healthy volunteers.

The impedance is the sum of R and Ra. These values are derived from the impedance Ohm law stating that when an electrical current passes through the human body, the voltage’s difference between 2 points is proportional to the impedance of this body [13].

One month after admission to the ED, all patients’ medical records, therapeutic management, and laboratory results (all biology, BNP, cardiac ultrasound, chest radiograph, and any other examination performed during stay in ECC or in subsequent hospitalization) were reviewed by 2 independent cardiology experts blinded to bioimpedance measurements.

Clinical and laboratory data in medical records were reviewed to es- tablish the etiologic diagnosis of dyspnea. Patients were then classified into the AHF or non-AHF group. This expert classification set the refer- ence for our study population.

Statistical analysis

Qualitative variables are expressed as percentages or absolute numerical value, whereas continuous variables are expressed as mean +- standard deviation or median and interquartile range for non-Gaussian distribution. Resistance and Ra are expressed in ohms (?); total and extracellular water, in liters; and BNP, in nanograms per liter.

The comparison between groups was performed by a nonparametric Wilcoxon rank sum test for quantitative variables and a ?² or Fisher test for qualitative variables. The various data measured by bioimpedance were related to final discharge diagnosis of heart failure by univariate logistic regression. We performed multivariate logistic regression models to assess the effect of prespecified potential confounders such as age or renal failure. The first model integrated each bioimpedance vector analysis (BIVA) parameter separately adjusting for age. The second model integrated each BIVA parameter adjusting for age and glomerular filtration rate (GFR) value at admission. The relationship between BNP levels and each bioimpedance parameter was assessed by linear regression. The relationship between different LVEDP categories and each bioimpedance parameter was assessed by ordinal logistic regression.

In the multivariable analysis, the association between BIVA and heart failure was analyzed using 3 models with different covariate ad- justments: model 1 included age, model 2 included age and GFR, and model 3 included model 2 plus BNP levels.

We tested and compared the discrimination ability of BIVA and BNP for heart failure in the univariate and multivariate analyses beyond through differences in area under the curve (AUC) derived from receiv- er operating characteristic (ROC) analysis. The AUCs were computed based on the predicted risk from multivariate modeling and compared using a parametric method. Comparisons between ROC curves obtained with different models were performed with C statistics. Tests were con- sidered significant with a P value b .05. All statistical measurements were performed using SPSS 20.0 (Chicago, IL) software and GraphPad Software Version 6.0 (San Diego, CA).

1. Results
   1. Study population characteristics

There were 77 patients, with a mean age of 68 +- 14 years. Thirty- eight (49.4%) women were included in the study. The principal patient characteristics at presentation are presented in Table 1 according to heart failure status. The diagnosis of AHF was confirmed for 37 (48%) patients of the study population.

BIVA measurements

The respective values of Ra, R, TBW, and EEW in the population study are presented in Table 2. Of the 4 parameters assessed by BIVA,

Table 1

Principal characteristics of study population

	AHF (n = 37)	No AHF (n = 40)	P value
Age (y)	71.5 +- 13.6	64 +- 14.5	.034
Male sex, n (%)	20 (54)	18 (45)	.49
Diabetes mellitus, n (%)	11 (30.55)	13(33.33)	.49
Hypertension, n (%)	21 (58.33)	18 (46.15)	.20
History of cardiac disease, n (%)	7 (19.44)	13 (33.33)	.13
History of pulmonary disease, n (%)	19 (52.77)	7 (17.95)	.002
CRP (mg/L)	25.3 +- 32.6	27 +- 48.7	.39
GFR (mL/mn)	65.9 +- 46.5	86.1 +- 41.0	.007
Hemoglobin (g/dL)	12.4 +- 1.8	13.4 +- 1.7	.017
LVEF <= 45%, n (%) LVEDP, n (%)	21 (57)	8 (20)	.002 <=.001
Low	3 (9.7)	28 (98.23)
Intermediate	9 (45)	11 (55)
High	25 (96.2)	1 (3.8)

All values are expressed as the means +- SDs or absolute numbers and percentage. CRP: C- reactive protein; GFR: glomerular filtration rate; LVEF: left ventricular ejection fraction. Low LVEDP: E/e’ <= 8; intermediate LVEDP: E/e' 9-14; and high LVEDP: E/e' >= 15.

Fig. 1. Relationship between BNP and Ra. Whisker plots of BNP according to different Ra intervals. Interval 1: b39 ?; interval 2: >=39. ***P b .001.

only Ra had a significant relationship with the final diagnosis of AHF (odds ratio [OR] = 0.9; 95% confidence interval [CI], 5.120.4; P b .001). The average Ra was 45.2 +- 19.3 ? in patients with extracardiac dyspnea and 32.7 +- 14.3 ? for patients with AHF (P b .05). The area under the ROC curve of the Ra was 0.76. Other parameters measured by BIVA (R, TBW, and EEW) showed no significant differences between patients with and without AHF (Table 1).

An analysis of the healthy group was performed. There were no significant differences between the healthy controls and AHF patients for R, TBW, and EEW. Conversely, Ra was significantly decreased in AHF patients compared to healthy controls (32.7 +- 14.3 vs 47.4 +- 12.4, respectively; P = .001).

Relationship and comparison with BNP levels and ROC curves. Con- founding factors

There was a significant inverse linear relationship between BNP levels and Ra values (r = -0.47, P b .001) (Fig. 1).

Echocardiographic indices of high LVEDP were also associated signif- icantly with AHF (OR = 81.25; 95% CI, 9.9-664.4; P b .0001). The differ- ent echocardiographic LVEDP categories were significantly inversely related to Ra values as shown in Fig. 2. Twenty patients (26%) presented E/e’ ratio between 9 and 14 (9 [12%] of them in AHF).

Confouding factors.

We studied BIVA parameters with heart failure diagnosis depending of age, GFR, and BNP levels (model 1, model 2, and model 3) with logis- tic regression. Only Ra shows significative results (TBW, EBW, and R; P = not significant). In model 1, age did not affect Ra results (OR = 0.95; CI, 0.91-0.99; P = .01). In model 2, age and GFR did not affect Ra results (OR = 0.95; CI, 0.91-0.99; P = .01). In model 3, BNP affected Ra significance (Ra P = .69; BNP OR = 1; CI, 1.004-1.01; P <= .001).

Brain natriuretic peptide is directly correlated with heart failure diagno- sis (cf Fig. 3).

The addition of Ra values to BNP levels in a multivariable model for AHF accurate diagnosis did not significantly improve the diagnostic per- formance of BNP levels alone by C statistic analysis (P = not significant).

Table 2

BIVA parameters according to patient group

	AHF	No AHF	Control group	P	95% CI
Ra (?)	32.7 +- 14.3	45.4 +- 19.3	47.4 +- 12.4	b.001	5.1 to 20.4
R (?)	425.8 +- 112	409.5 +- 91.7	439 +- 46.6	.83	-63 to 30.5
TBW(L)	36.7 +- 9.6	37.4 +- 8.4	34.4 +- 6	.69	-3.5 to 4.8
EEW(L)	15.6 +- 4.2	14.8 +- 3.1	14.1 +- 2	.42	-2.5 to 0.9
BNP (ng/L)	1050.3 +- 989	148.7 +- 181.1	NA	b.001	-1240.6 to

-562.5

Serum BNP levels were significantly higher among patients with AHF compared to patients with noncardiac dyspnea (1050.3 +- 989 vs 148.7 +- 181.1 ng/L; P b .001). The relationship between BNP values and AHF final diagnosis was highly significant with an AUC (ROC) of

0.92 (Fig. 4).

A Ra threshold of 39 ? yielded a sensitivity of 67%, a specificity of 76%, a positive predictive value of 58%, and a negative predictive value of 75% for the diagnosis of AHF. The AUC ROC was 0.76 (Fig. 5).

1. Discussion

In our study, our principal findings were as follows:

1. Reactance was significantly related to clinical, ultrasound, and biomarker indices of AHF in a patient population presenting to the ED for dyspnea.
2. There was no significant association of AHF status with all other BIVA parameters (EBW, R, and TBW).
3. Bioimpedance vector analysis did not significantly improve the diagnostic performance for AHF in addition to BNP alone.

Reactance represents the property of a material to oppose an electric current. It depends on the electrical current intensity. Resistance, on the other hand, depends on the current’s frequency. Both are used to ex- trapolate other BIVA variables such as TBW, EBW, and fat mass. Several reports have shown that bioimpedance measurements were useful in various health areas such as nutrition to assess the body’s composition

Fig. 2. Relationship of Ra and ultrasound indices of LVEDP. Whisker plots of Ra and cardiac ultrasound indices of LVEDP. The LVEDPs are divided into 3 categories: low LVEDP, E/e’ <= 8; intermediate LVEDP, E/e' 9-14 ; and high LVEDP, E/e' >= 15. *P b .05; ***P b .001.

Fig. 3. Receiving operator characteristic curves areas for heart failure diagnosis comparing 3 different models and BNP alone. Model 1: Ra and age; model 2: Ra, age, and GFR; and model 3: Ra, age, GFR, and BNP.

[14,26]. This noninvasive, easily available, and inexpensive technology is interesting in the setting of AHF diagnosis. The fluid overload in AHF patients induces variations in electricity conduction due to water tissue content variations. It is expected that Ra and R decrease, whereas TBW and EEW increase, in AHF patients.

Our results show a significant relationship between the Ra measured with BIVA and AHF diagnosis. In patients presenting with AHF, Ra was significantly lower compared to patients with non-AHF dyspnea and healthy volunteers. In AHF status, the electric current is faster in an in- flated body with less R. We also compared our results with BNP levels that are significantly related to AHF [8,27]. The BNP levels significantly correlated with Ra. We found a significant inverse relationship between Ra and BNP levels as well as ultrasound indices of LVEDP. Bioimpedance vector analysis can also be interesting when cardiac LVEDP indices are inconclusive or not assessable.

In our study, Ra had a sensitivity of 67% and a specificity of 76% with a threshold of 38.85 ? for AHF diagnosis. In a previous report, Parrinello et al [28] found similar results for BIVA’s diagnostic performance. In their study, Ra and R were significantly decreased in patients with AHF; and the AUC of R was 0.93 with sensitivity of 81% and specificity of 94%. In our study, only Ra was significantly related to AHF with lower sensitivity and specificity. These differences may be explained in part by different inclusion criteria and study populations but also with use of a different device. Parrinello et al excluded patients with

Fig. 4. Receiving operator characteristic of BNP. Area under curve: 0.92.

Fig. 5. Receiving operator characteristic of Ra. Area under curve: 0.76.

acute coronary syndromes, recent heart surgery, or clinical conditions of high hydration such as cirrhotic or chronic renal failure. The exclusion of patients who may have a lower impedance (fluid overload with extraCardiac origin) could result in a substantial selection bias in their study. Moreover, we used a different technology–and therefore differ- ent algorithms and calibrations–than in their study.

In another setting, Springfield et al [22] found that thoracic imped- ance had a sensitivity of 92 % and a sensitivity of 88% for the diagnosis of AHF. These results are not directly comparable to ours because tho- racic impedance mainly studies cardiac output and aortic velocity.

In our study R, EBW, and TBW were not significantly related to AHF. These negative results may be partly explained by our BIVA device tech- nology. Many algorithms that are reported from measured values are obtained from specific manufacturer calculations [29,30]. Extracellular body water and TBW are parameters measured upon these algorithms that might apply to certain conditions, but not AHF. Our results are con- sistent with reviews and reports where Ra and R appear to be the most reliable parameters related to AHF [20,24,28]. Reactance and R repre- sent raw electric data saved by BIVA. These results are variable and de- pend of the device used. The Ra is linked to the current intensity, whereas the R relates to its frequency. We used a multifrequency meter in our study that delivers several current frequencies in a few sec- onds. We studied the global value of R (the infinite R) that is an average of the tested frequencies. Our hypothesis is that a particular frequency–and thus particular R–may be significantly associated with a fluid overload and the infinite R.

Bioimpedance vector analysis lacks sensitivity for the detection of compartmented edema such as pleural or Abdominal effusion [15,24]. This is explained by the fact that electric current is well conducted in the interstitial space but penetrates poorly into cavities such as the ab- dominal or pericardial cavities. In our study, this might participate in the lower diagnostic performance for AHF by BIVA. Nevertheless, no pa- tients with this particular case in our study population were reported.

Bioimpedance could be interesting in clinical situations where BNP levels are inconclusive. The BNP thresholds are not standardized and depend on manufacturer and laboratories settings. In the literature, a value greater than 100 ng/L often indicates heart failure. However, BNP levels vary according to physiological and pathological conditions independent of heart failure (age, obesity, pulmonary embolism, or pulmonary infection). These situations that can be found in patients presenting to the ED with dyspnea symptoms affect BNP levels specific- ity. Furthermore, in patients with CHF, BNP levels can be elevated; and it is harder to determine whether dyspnea symptoms are related to HF exacerbation or another extracardiac factor. In such cases, BIVA measurement could offer a complementary means for AHF diagnosis in the ED where cardiac ultrasound scan is not readily available as

Di Somma et al [31] tried to show. In their study, BIVA did not increase diagnostic accuracy provided by BNP. But in patients with BNP in “gray zones,” BIVA showed a significant addictive improvement for heart failure diagnosis.

Bioimpedance vector analysis could also have an interest during the follow-up of CHF patients. Bioimpedance vector analysis could also be of help in the management and monitoring of diuretic prescriptions for decompensated CHF patients. Bioimpedance vector analysis could help physicians to detect early fluid changes and modify their therapeu- tic management. Ideally, BIVA could be implemented at home in select- ed CHF patients to monitor and tailor therapeutic adjustments. This feature has already been implemented in new-generation pacemakers or Implantable cardiac defibrillators by several vendors. However, the use of these features remains to be tested in clinical studies on hard clin- ical outcomes in large populations of patients.

Study limitations

Our study has several limitations. First of all, our study sample is small, which increases the risk of potential hazardous findings. Second, our hospital is a specialized hospital in cardiac and pulmonary diseases. Patients consulting to our ECC ward might not reflect as accurately a pa- tient population consulting for dyspnea in a general ED. Other than this potential bias, selection parameters were broad; therefore, we think that our findings are still representative of a general population consult- ing for these symptoms. Third, the final expert diagnosis of heart failure was not blinded to BNP levels. This probably has an effect on the results for BNP’s discrimination of heart failure. However, the purpose of this study was to assess the relationship of BIVA parameters with AHF diag- nosis in an emergency setting and not to reassess BNP that is now a well-validated Diagnostic biomarker used worldwide for the routine di- agnosis and management of heart failure patients. Finally, the absence of relationship of 3 out of 4 BIVA parameters with AHF status might be explained by a statistical lack of power but also by independent algo- rithm manufacturer settings, which are calibrated on healthy subjects and might not be fitted for patients presenting with diseased states.

1. Conclusions

In a population of patients presenting to the ED with dyspnea, BIVA was significantly related to the diagnosis of AHF but did not improve the diagnostic performance for AHF in addition to BNP alone. Larger clinical studies comparing various BIVA technologies in large samples of patients and different clinical settings are warranted to assess the place of BIVA in heart failure patient’s management.

References

1. Watson RD, Gibbs CR, Lipp GY. ABC of heart failure: clinical features and complica- tions. Br J Med 2000;320:236-9.
2. Gheorgiade M, Filippatos G, De Luca L, Burnett J. Congestion in acute heart failure syn- dromes: an essential target of evaluation and treatment. Am J Med 2006;119:S3-S10.
3. McMurray John JV, Adamopoulos Stamatis, Anker Stefan D, Auricchio Angelo, Bohm Michael, Dickstein Kenneth, et al. ESC guidelines on acute heart failure. Eur Heart J 2012;33:1787-847.
4. Adams Jr Kirkwood F, Fonarow Gregg C, Emerman Charles L, LeJemtel Thierry H, Costanzo Maria Rosa, Abraham William T, et al. Characteristics and outcomes of pa-

tients hospitalized for heart failure in the United States: rationale, design, and pre- liminary observations from the first 100,000 cases in the Acute Decompensated Heart Failure National Registry (ADHERE). Am Heart J 2005;149:209-16.

1. Zannad F, Adamopoulos C, Mebazaa A, Gheorghiade M. The challenge of acute de- compensated heart failure. Heart Fail Rev 2006;11(2):135-9.
2. Peacock WF, Allegra J, Ander D, Collins S, Diercks D, Emerman C, et al. Management of acute decompensated heart failure in the emergency department. Congest Heart Fail 2003;9(Suppl. s5):7-18.
3. Fonseca C. Diagnosis of heart failure in primary care. Heart Fail Rev 2006;11:95-107.
4. Tang WHW, Girod JP, Lee MJ, Starling RC, Young JB, Van Lente F, et al. Plasma B type natriuretic peptide levels in ambulatory patients with established chronic symptom- atic systolic heart failure. Circulation 2003;108:2950-3.
5. Maisel AS, Krishnaswamy P, Nowak RM, McCord J, Hollander JE, Duc P, et al. Rapid measurement of B-type natriuretic peptide in the emergency diagnosis of heart fail- ure. N Engl J Med 2002;347:161-7.
6. Packer M. Should B-Type natriuretic peptide be measured routinely to guide the di- agnosis and management of chronic heart failure? Circulation 2003;108:2950-3.
7. Knudsen CW, Omland T, Clopton P, Westheim A, Abraham WT, Storrow AB, et al. Di- agnostic value of B-type natriuretic peptide and chest radiographic findings in pa- tients with acute dyspnea. Am J Med 2004;116(6):363-8.
8. Squara P, Denjean D, Estagnasie P, Brusset A, Dib JC, Dubois C. Noninvasive cardiac output monitoring: a clinical validation. Crit Care 2007;11:289-490.
9. Abraham WT. Intrathoracic impedance monitoring for early detection of impending heart failure decompensation. Congest Heart Fail 2007;13:113-5.
10. Martinez L, Ramirez CE, Tejeda A, Lafuente E, Rosales LP, Gonzalez V, et al. Bioelec- trical impedance and strength measurements in patients with heart failure: compar- ison with functional class. Nutrition 2007;23:412-8.
11. Piccoli A, Rossi B, Pillon L, Bucciante G. A new method for monitoring body fluid var- iation by bioimpedance analysis. The RXc graph. Kidney Int 1994;46:534-9.
12. Panella C, Guglielmi FW, Mastronuzzi T, Francavilla A. Whole-body and segmental bioelectrical parameters in Chronic liver disease: effect of gender and disease stages. Hepatology 1995;21:352-8.
13. Torres D, Parrinello G, Paterna S, Di Pasquale P, Torres A, Trapanese C, et al. A new option in measuring bioimpedance in congest heart failure. Am Heart J 2009; 158(1):e1.
14. Peackok W, Albert NM, Kies P, White RD, Emerman C. Bioimpedance monitoring: better than x-ray for predicting abnormal pulmonary fluid? Congest Heart Fail 2000;6:86-9.
15. Maisch S, Bohm S, Sola J, Goepfert M, Kubitz JC, Richter HP, et al. Heart-lung interac- tions measured by electrical impedance tomography. Crit Care 2011;39:2173-6.
16. Tang WHW, Tong W. Measuring impedance in congestive heart failure: current op- tions and clinical applications. Am Heart J 2009;157:402-11.
17. Peacok WF, Summers R, Emerman C. Impact of impedance cardiography on diagno- sis and therapy of emergent dyspnea: the ED-IMPACT trial. Acad Emerg Med 2006; 13(4):365-71.
18. Springfield CL, Sebat F, Johnson D, Lengle S, Sebat C. Utility of impedance cardiogra- phy to determine cardiac versus noncardiac cause of dyspnea in the emergency de- partment. Congest Heart Fail 2004;10:14-6.
19. Perego GB, Landolina M, Vergara G, Lunati M, Zanotto G, et al. Implantable CRT de- vice diagnostics identify patients with increased risk for heart failure hopsitalization. J Interv Card Electrophysiol 2008;23(3):235-42.
20. Yu CM, Wang L, Chau E, Chan RH, Kong SL, Tang MO. Intrathoracic impedance mon- itoring in patients with heart failure: correlation with fluid status and feasibility of early warning preceding hospitalization. Circulation 2005;112:841-8.
21. Nagueh SF, Appleton CP, Gillebert TC, Marino PN, Oh JK, Smiseth OA, et al. Recom- mendations for the evaluation of left ventricular Diastolic function by echocardiogra- phy. J Am Soc Echocardiogr 2009;22(2):107-33.
22. Moon JR. Body composition in athletes and sports nutrition: an examination of the bioimpedance analysis technique. Eur J Clin Nutr 2013;67(Suppl. 1):S54-9.
23. Braunwald E. Biomarkers in heart failure. N Engl J Med 2008;358:2148-59.
24. Parrinello G, Paterna S, Di Pasquale P, Torres D, Fatta A, Mezzero M, et al. The useful- ness of bioelectrical impedance analysis in differentiating dyspnea due to decom- pensated heart failure. J Card Fail 2008;14:676-86.
25. Kubicek WG, Karnegis JN, Patterson RP. Development and evaluation of an imped- ance cardiac output system. Aerosp Med 1966;37:1208-12.
26. Raaijmarkers E, Faes TJC, Goovaerts HG, de Vries PM, Heetraat RM. The inaccuracy of Kubicek’one-cylinder model in thoracic impedance cardiography. IEEE Trans Biomed Eng 1997;44:70-6.
27. Di Somma S, Lalle I, Magrini L, Russo V, Navarin S, Castello L, et al. Additive diagnos- tic and prognostic value of bioelectrical impedance vector analysis (BIVA) to brain natriuretic peptide ‘grey-zone’ in patients with acute heart failure in the emergency department. Eur Heart J Acute Cardiovasc Care 2014;3(2):167-75.