The current evaluation of acute heart failure (HF) does not allow an adequate prediction of its evolution. The electrical bioimpedance (BI) allows knowing the state of blood volume, until now only with fixed equipment. We have developed and validated a portable and wireless device to measure BI at the ankle (IVOL). The objective of the study is to know the long-term prognostic value of the point measurement of BI with IVOL in patients with acute HF.
MethodsA prospective cohort study of unselected patients admitted for acute HF in a tertiary hospital. The association between BI and different clinical, analytical and echocardiographic variables on admission and clinical evolution were analyzed.
Results76 patients were included (mean age 66.1 years, 71.1% men, 68.4% hypertensive, 34.2% diabetic, mean NT-ProBNP: 7,103 pg/ml). Of these, 52.6% with non-preserved left ventricular ejection fraction (LVEF) (<50%) and 56.6% with right ventricular (RV) dysfunction. 26.3% died during a mean follow-up of 35.8 months. Survival in patients with BI ≤ 21,8 Ω was lower, globally and in the subgroups of patients without preserved LVEF and with RV dysfunction, p < 0.008). In the multivariate analysis, a BI ≥ 21.8 Ω was an independent survival factor (HR 0.242, 95% CI:0.86−0.681, p = 0.007).
ConclusionsBI values measured with IVOL may be an independent predictor of long-term mortality in patients hospitalized for acute HF. This prognostic value is maintained in patients without preserved LVEF function and with RV dysfunction.
La evaluación actual de la insuficiencia cardiaca (IC) aguda no permite predecir adecuadamente su evolución. La bioimpedancia eléctrica (BI) permite conocer el estado de volemia, hasta ahora solo con equipos fijos. Hemos desarrollado y validado un dispositivo portátil e inalámbrico para medir BI en el tobillo (IVOL). El objetivo del estudio es conocer el valor pronóstico a largo plazo de la medición puntual de la BI con IVOL en pacientes con IC aguda.
MétodosEstudio de cohorte prospectivo de pacientes no seleccionados ingresados por IC aguda en un hospital de tercer nivel. Se analizó la asociación entre BI y diferentes variables (clínicas, analíticas y ecocardiográficos) al ingreso y evolución clínica.
ResultadosSe incluyeron 76 pacientes (edad media 66,1 años, 71,1% varones, 68,4% hipertensos, 34,2% diabéticos, NT-ProBNP medio: 7.103 pg/mL). De ellos, 52,6% con fracción de eyección del ventrículo izquierdo (FEVI) no preservada (<50%) y 56,6% con disfunción del ventrículo derecho (VD). 26,3% fallecieron durante un seguimiento medio de 35,8 meses. La supervivencia en pacientes con BI ≤ 21,8 Ω fue menor, globalmente, y en los subgrupos de pacientes sin FEVI preservada y con disfunción del VD, p < 0,008). En el análisis multivariante una BI ≥21,8 Ω fue un factor independiente de supervivencia (HR 0,242, IC95%:0,86−0,681, p = 0,007).
ConclusionesLos valores de BI medidos con IVOL pueden ser un predictor independiente de mortalidad a largo plazo en pacientes hospitalizados por IC aguda. Este valor pronóstico se mantiene en pacientes con FEVI no preservada y con disfunción ventricular derecha.
The natural history of heart failure (HF) is characterised by episodes of acute HF due to the rapid onset or worsening of symptoms and/or signs of known chronic HF or as the first manifestation of the syndrome.1
It is the leading cause of hospitalisation in Spain in people over 65 years of age, accounting for 3% of all hospital admissions, 2.5% of healthcare costs2 and 51% of the average cost of HF patients, estimated at more than ;15,000 per year.3 On the other hand, hospitalisation episodes continue to have a major impact on the course of the disease, with in-hospital mortality in Spain currently at 9.4% and a readmission rate of 9.7%,4 and an annual mortality rate of 23.6% in recent registries in European and Mediterranean countries.5
This is at least partly because standard HF assessment (monitoring of vital signs, weight, body mass index, urine output, electrocardiogram [ECG], imaging tests and natriuretic peptide measurement),1 does not reliably predict HF decompensation, nor does it accurately monitor resolution of volume overload during admission or clinical outcome after discharge.6
Previous studies have shown that electrical bioimpedance (BI) spectroscopy can measure blood volume, especially in renal failure7 and HF.8 In HF, most studies have used invasive intrathoracic devices to detect pulmonary congestion,9,10 and non-invasive commercial global blood volume measurement devices to detect systemic congestion, which have usability issues. A large study has shown that leg BI measurement allows early diagnosis of HF in the general population,11 and some other studies have preliminarily shown that the BI values measured during admission are predictive of short-term outcome in acute HF.12,13
We have developed a low-cost, wireless, portable BI device to measure the BI spectrum at the ankle (IVOL device), which has been validated in a pilot study against a high-performance commercial device.14 The objective of this study is to determine the long-term prognostic value of BI measurements using an IVOL portable device in a cohort of patients hospitalised with acute HF.
MethodsPatientsUnselected adult patients hospitalised for acute HF at the Hospital U. Virgen del Rocío in Seville between 1 March 2015 and 20 April 2017 were prospectively included. Inclusion criteria were: (a) adults ≥ 18 years admitted for acute HF with confirmed diagnosis according to Framingham criteria15 and (b) written informed consent. Exclusion criteria were: (a) coronary artery disease, (b) oedema secondary to venous disease and/or lymphoedema and (c) lower limb amputation, ulcers or burns. On admission, transthoracic echocardiography and BI measurement was performed before starting treatment with the IVOL device.
A standardised electronic data form was developed which prospectively included during admission: demographic data, comorbidities, degree of oedema16 in the right lower limb, echocardiographic parameters (LV ejection fraction [LVEF], tricuspid annular plane systolic excursion [TAPSE] and pulmonary artery pressure [PAP]), biological (plasma haemoglobin levels, serum sodium and NT-pro-BNP levels) and diuretic treatment administered (drug and daily dose).
The primary outcome variable was all-cause mortality, and the secondary variables were readmissions to hospital or the emergency department for HF exacerbation. Outcome assessment was performed between 25 and 30 June 2021, by consulting the Andalusian Digital Health Record (Historia Digital de Salud de Andalucía-Diraya). The date and status of the last available data were recorded.
Electrical bioimpedance measurement systemThe IVOL measurement equipment (Fig. 1)14 is based on a 2-wire impedance measurement device (AD5933), to which an analogue electronic unit is added to switch to a 4-wire measurement ("front-end"), a microprocessor and a "Bluetooth" communication module. Data management and display of results are carried out on a mobile phone with an Android® operating system. The designed system has a dynamic range of up to 300 ohms (Ω), with a resolution of 0.1 Ω. The main specifications of the device are: measurement frequencies: 5, 10, 20, 20, 50, 100 and 200 kHz; dynamic range: 10–300; resolution of 0.1 Ω; accuracy: ± 5% (from 10 to 100 Ω); phase resolution: 0.01 degrees; measurement time: 8 s for the 6 frequencies.
Measurement protocol
Localised BI measurement was obtained on the right lower limb with 4 commercial ECG electrodes, with a distance between electrode centres of 6 and 3.5 cm between current injectors and voltage detectors (Fig. 1a and b). Current injection was performed longitudinally to the leg (L-spacing). Voltage measurements were made on electrodes placed in parallel at a D distance and attached to the patient in the position shown in Fig. 1. BI measurements were taken in the morning with the patient bedridden. The impedance value used in this study is 50 kHz and it is shown in the tables as the IVOL BI value.
Statistical analysisQualitative variables were expressed as percentages and continuous variables as means and standard deviation (SD). Differences in qualitative variables were analysed using the Chi-square test or Fisher's exact test when the expected frequency was less than 5. Quantitative variables were analysed using Student's t-test, after confirmation of normal data distribution using the Shapiro–Wilk test. Non-normal distribution data were analysed using the Mann–Whitney U test.
On the other hand, to obtain the cut-off value of BI, a sensitivity analysis was performed using Receiver Operating Characteristics (ROC) curves and, subsequently, mortality was compared according to the values obtained for BI using Kaplan-Meier curves (Mantel-Cox test). Finally, a Cox multivariate analysis was performed, including variables with statistical association in the bivariate analysis. Hazard ratios (HR) with their 95% confidence interval (95% CI) were calculated. The analysis was performed with the IBM SPSS® Statistics 21 software (IBM Corporation, Armonk, NY, USA).
ResultsA total of 76 patients were included with a mean age of 66.07 ± 13.09 years, 71.1% were male. Of these, 52.6% had HF with non-preserved LVEF (< 50%), and 56.6% had right ventricular (RV) dysfunction. The remaining characteristics of the patients evaluated are shown in Table 1. The patient’s mean follow-up was 36.83 months. During the follow-up 18.4% were readmitted to hospital and 26.3% died.
Baseline characteristics of the study population, overall and according to outcome.
| Variables | Overall (n = 76) | According to outcome: death | P-value | |
|---|---|---|---|---|
| No (n = 56) | Yes (n = 20) | |||
| Demographic | ||||
| Age (years), mean ± SD | 66.07 ± 13.09 | 53.82 ± 12.48 | 72.35 ± 9.71 | 0.011* |
| Males, n (%) | 54 (71.1) | 36 (64.3) | 18 (90) | 0.043** |
| Comorbidity | ||||
| Hypertension, n (%) | 52 (68.4) | 36 (64.2) | 16 (80) | 0.226*** |
| Diabetes mellitus, n (%) | 26 (34.2) | 16 (28.6) | 10 (50) | 0.103** |
| Dyslipidemia, n (%) | 38 (50) | 27 (48.2) | 11 (55) | 0.795** |
| BMI (kg/m2) > 30, n (%) | 32 (42.1) | 24 (42.8) | 8 (40) | 0.787** |
| Chronic kidney disease, n (%) | 25 (32.9) | 15 (26.8) | 10 (50) | 0.160** |
| Pacemaker carrier, n (%) | 11 (14.5) | 7 (12.5) | 4 (20) | 0.446*** |
| Atrial fibrillation, n (%) | 40 (52.6) | 27 (48.2) | 13 (65) | 0.197** |
| Echocardiographic data | ||||
| LVEF < 50%, n (%) | 40 (52.6) | 26 (46.4) | 14 (70) | 0.07** |
| TAPSE < 17 mmHg, n (%) | 43 (56.6) | 32 (58.2) | 11 (55) | 0.805** |
| PAP > 40 mmHg, n (%) | 31 (40.8) | 21 (38.2) | 10 (50) | 0.358** |
| Clinical signs | ||||
| Oedemas, n (%) | ||||
| No | 23 (30.3) | 23 (30.3) | 0 (0) | < 0.0001*** |
| Grade 1 | 15 (19.7) | 11 (19.7) | 4 (20) | 0.004*** |
| Grade 2 | 19 (25) | 11 (25) | 8 (40) | |
| Grade 3 | 19 (25) | 11 (25) | 8 (40) | |
| Biological parameters | ||||
| BI (Ohms), mean ± SD | 24.3 ± 8.5 | 19 ± 5 | 0.011**** | |
| Na (mEq/l), mean ± SD | 136 ± 6.6 | 137.40 ± 4.5 | 132.1 ± 9.5 | 0.013**** |
| Haemoglobin (mg/dl), mean ± SD | 11.03 ± 1.9 | 10.89 ± 1.8 | 11.41 ± 2.1 | 0.313* |
| Nt-pro-BNP (pg/mL), mean ± SD | 7,103.8 ± 8,931.5a | 5,669.1 ± 9,694.8b | 9,176.2 ± 7,761.7c | 0.03**** |
| Treatment | ||||
| Treated with furosemide, n (%) | 53 (69.7) | 37 (66.1) | 16 (80) | 0.395*** |
| Dose of furosemide (mg/d) | 64.81 ± 29.67 | 54 ± 33.95 | 71.18 ± 31.2 | 0.031**** |
| Re-admissions | ||||
| HF readmissions to the ward, n (%) | 14 (18.4) | 5 (8.9) | 9 (45) | <0.0001** |
| HF readmissions to the ED, n (%) | 13 (17.1) | 6 (10.7) | 7 (35) | 0.033** |
BI, electrical bioimpedance measured with IVOL; BMI, body mass index; HF, heart failure; LVEF, left ventricular ejection fraction; Na, plasma sodium; Nt-pro-BNP, N-terminal pro-B-type natriuretic peptide; PAP, pulmonary artery pressure; TAPSE, tricuspid annular plane systolic excursion; SD, standard deviation.
Among patients who died, the proportion of males, mean age, percentage of subjects with oedema and mean Nt-pro-BNP levels were higher, while plasma sodium and BI values measured by IVOL were lower (Table 1).
When analysing patient changes according to LVEF (Table 2), those who died in the subgroup with LVEF < 50% had a higher proportion of oedema and lower BI values on admission, while the subgroup with preserved LVEF (≥ 50%) had a higher mean age and higher plasma NT-pro-BNP levels and lower BI values.
Baseline characteristics and outcome according to LVEF.
| Variable | LVEF < 50% (n = 40) | LVEF ≥ 50% (n = 36) | ||||
|---|---|---|---|---|---|---|
| Death | P-value | Death | P-value | |||
| No (n = 26) | Yes (n = 14) | No (n = 30) | Yes (n = 6) | |||
| Demographic | ||||||
| Age, average ± SD | 62.23 ± 13.25 | 71.36 ± 8.45 | 0.227*** | 62.60 ± 13.79 | 74.67 ± 12.79 | 0.056* |
| Males, n (%) | 18 (69.2) | 13 (92.9) | 0.91** | 18 (60) | 5 (83.4) | 0.276** |
| Echocardiographic data | ||||||
| TAPSE < 17 mmHg, n (%) | 18 (72) | 11 (78.6) | 0.48** | 14 (46.7) | 0 (0) | 0.038** |
| PAP > 40 mmHg, n (%) | 8 (32) | 7 (50) | 0.268**** | 13 (43.3) | 3 (50) | 0.554** |
| Clinical signs | ||||||
| Oedemas, n (%) | ||||||
| No | 13 (50) | 0 (0) | 0.001** | 10 (33.3) | 0 (0) | 0.118** |
| Grade 1 | 5 (19.2) | 3 (21.4) | 0.007** | 6 (20) | 1 (16.7) | 0.272** |
| Grade 2 | 5 (19.2) | 5 (35.7) | 6 (20) | 3 (50) | ||
| Grade 3 | 3 (11.5) | 6 (42.9) | 8 (27.7) | 2 (33.3) | ||
| Biological parameters | ||||||
| BI (Ohms), mean ± SD | 25.13 ± 9.3 | 19.53 ± 4.9 | 0.04*** | 23.56 ± 7.81 | 17.6633 ± 5.49 | 0.088* |
| Na (mEq/l), mean ± SD | 137.29 ± 5.61 | 131.08 ± 10.37 | 0.944* | 137.38 ± 3.53 | 134.4 ± 7.43 | 0.335*** |
| Haemoglobin (mg/dl), mean ± SD | 11.38 ± 1.86 | 11.43 ± 2.16 | 0.944* | 10.52 ± 1.7 | 11.37 ± 2.31 | 0.335* |
| Treatment | ||||||
| Dose of furosemide (mg/d) | 49.33 ± 30.11 | 66.92 ± 35.5 | 0.149* | 56.80 ± 37.7 | 85 ± 25.17 | 0.150* |
| Re-admissions | ||||||
| HF readmissions to the ward, n (%) | 3 (11.5) | 8 (57.1) | 0.004** | 4 (13.3) | 5 (83.3) | 0.034** |
| HF readmissions to the ED, n (%) | 3 (11.5) | 4 (28.6) | 0.161** | 3 (10) | 3 (50) | 0.046** |
BI, electrical bioimpedance measured with IVOL; BMI, body mass index; HF, heart failure; LVEF, left ventricular ejection fraction; Na, plasma sodium; Nt-pro-BNP, N-terminal pro-B-type natriuretic peptide; PAP, pulmonary artery pressure; TAPSE, tricuspid annular plane systolic excursion; SD, standard deviation.
When analysing patient changes according to RV dysfunction (Table 3), among those who died in the subgroup with TAPSE < 17 mmHg, the proportion of subjects with left ventricular dysfunction was higher and BI values lower, while the subgroup with TAPSE ≥ 17 mmHg had lower plasma sodium levels, with no significant differences in BI values.
Baseline characteristics and outcome according to RV TAPSE.
| Variable | TAPSE < 17 mmHg (n = 43) | P-value | TAPSE ≥ 17 mmHg (n = 32) | P-value | ||
|---|---|---|---|---|---|---|
| Death | Death | |||||
| No (n = 32) | Yes (n = 11) | No (n = 23) | Yes (n = 9) | |||
| Demographic | ||||||
| Age, average ± SD | 65.44 ± 14.08 | 73 ± 7.8 | 0.126a,d | 61.26 ± 12.72 | 71.56 ± 2.35 | 0.047b |
| Males, n (%) | 21 (65.6) | 221 (11) | 0.022 c | 14 (60.9) | 7 (77.8) | 0.318c |
| Echocardiographic data | ||||||
| LVEF < 50%, n (%) | 18 (56.3) | 11 (100) | 0.006c | 7 (30.4) | 3 (33.2) | 0.595c |
| PAP > 40 mmHg, n (%) | 13 (40.6) | 6 (54.5) | 0.432b | 8 (34.8) | 4 (44.4) | 0.456c |
| Clinical signs | ||||||
| Oedemas, n (%) | ||||||
| 14 (43.8) | 0 (0) | 0.006c | 8 (34.8) | 0 (0) | 0.047c | |
| 5 (15.6) | 1 (9.1) | 0.016c | 6 (26.1) | 3 (33.3) | 0.102c | |
| 8 (25) | 4 (36.4) | 3 (13) | 4 (44.4) | |||
| 5 (15.6) | 6 (54.6) | 6 (26.1) | 2 (22.2) | |||
| Biological parameters | ||||||
| BI (Ohms), mean ± SD | 24.8 ± 9.26 | 18.24 ± 3.21 | 0.027d | 23.86 ± 7.500 | 19.86 ± 6.75 | 0.181a |
| Na (mEq/l), mean ± SD | 136.4 ± 4.98 | 130.11 ± 11.91 | 0.092d | 138.45 ± 3.83 | 134.25 ± 5.8 | 0.028a |
| Haemoglobin (mg/dl), mean ± SD | 11.25 ± 1.98 | 10.76 ± 1.98 | 0.509a | 10.405 ± 1.52 | 12.13 ± 2.17 | 0.052d |
| Treatment | ||||||
| Dose of furosemide (mg/d) | 51 ± 22.92 | 73 ± 29.1 | 0.03a | 57 ± 42.69 | 68.57 ± 36.257 | 0.529a |
| Re-admissions | ||||||
| HF readmissions to the ward, n (%) | 4 (12.5) | 5 (45.4) | 0.034c | 1 (4.3) | 4 (44.4) | 0.01c |
| HF readmissions to ED, n (%) | 5 (15.6) | 3 (27.3) | 0.366c | 1 (4.4) | 4 (44.4) | 0.014c |
BI, electrical bioimpedance measured with IVOL; BMI, body mass index; HF, heart failure; LVEF, left ventricular ejection fraction; Na, plasma sodium; Nt-pro-BNP, N-terminal pro-B-type natriuretic peptide; PAP, pulmonary artery pressure; TAPSE, tricuspid annular plane systolic excursion; SD, standard deviation.
Survival curves dichotomized according to the BI value obtained in the ROC curves showed a shorter survival of patients with BI ≤ 21.80 Ω (42.07 vs. 56.57 months, log rank/Mantel-Cox; p = 0.017) (Fig. 2a). For the different subgroups (with or without LVEF ≥ 50% and with or without RV dysfunction), survival in patients with BI ≤ 21,80 Ω was lower only in the subgroups with LVEF < 50 (Fig. 2c) and TAPSE < 17 mmHg (Fig. 2e).
Finally, in the multivariate analysis, in which the variables age, sex, LVEF and BI were included, the only factors associated with mortality in the study population were age (HR: 1.067; 95% CI: 0.996–1.143; p = 0.064) and BI < 20 Ω (HR: 3.169; 95% CI: 1.001–9.96; p = 0.048). BI was an independent survival factor in the study population (BI ≥ 21.8 Ω; HR: 0.242; 95% CI: 0.86−0.681; p = 0.007) along with age and LVEF (Table 4).
Impact of different variables on survival. Cox multivariate analysis with adjusted hazard ratios.
| Variable | HR | 95% CI | P-value |
|---|---|---|---|
| Age (in years) | 1.093 | 1.024−1.167 | 0.007 |
| Sex (female vs. male) | 0.412 | 0.085−2.003 | 0.272 |
| LVEF (<50% vs. ≥50%) | 2.696 | 1.024−7.096 | 0.045 |
| IVOL (≥21,8 Ω vs. ≤21,8 Ω) | 0.242 | 0.86−0.681 | 0.007 |
95% CI, 95% confidence interval; HR, hazard ratio; IVOL, mean electrical bioimpedance using the IVOL device; LVEF, left ventricular ejection fraction; Ω, ohm.
The present study investigated the prognostic value of BI measurement in acute HF using a new portable device. This device, validated in a previous study against a commercial device for clinical use,14 is the first to allow wireless BI measurement of the leg at multiple frequencies and could therefore be used as a surrogate marker of congestion to detect subclinical congestion, in correlation or with other echocardiographic (e.g., cava diameter) and biological (e.g., CA-125) markers, and to guide early treatment.
Some findings should be highlighted with respect to the results obtained. First, the results indicate that BI is an independent prognostic marker of mortality in patients with HF, and that its prognostic value is maintained in patients with non-preserved LVEF (<50%), and in those with RV dysfunction (TAPSE < 17 mmHg). In this sense, low BI values (≤21.8 Ω) in the survival analysis are associated with significantly lower survival, both in the total set of subjects evaluated and in the subgroups of patients with non-preserved LVEF and with RV dysfunction.
In the bivariate analysis, mean BI values are lower in patients who die in the overall population and in the HF subgroups with non-preserved LVEF and RV diastolic dysfunction, although in the HF subgroup with preserved LVEF the differences are close to statistical significance (p = 0.088). The latter findings suggest that the predictive capacity of BI is lower in patients without right ventricular dysfunction, which is consistent with BI being a marker of the degree of peripheral oedema, which is less common in patients with preserved right ventricular function.17
Finally, in our study, the measurement of BI values using IVOL in hospitalised patients is an independent risk marker for long-term survival (mean follow-up: 35.8 months, range 0–60 months).
Another notable finding of the study is that during follow-up, 35.5% of patients were readmitted, 17.1% to the emergency department and 18.4% to the ward, suggesting that almost half of the cases could be resolved in the emergency department without the need for readmission to hospital.
To date, more than 150 prognostic models for HF have been described,18,19 with the strongest predictors being blood urea nitrogen and sodium,18 although a systematic review indicates that the studies used to develop them are subject to important biases19 because they have not been adequately calibrated,19,20 and a meta-analysis further indicates that the accuracy of the models for predicting mortality is moderate and even weaker for predicting hospitalization,18 as we have shown in a study on the calibration of prognostic scores.20
Clinical BI studies are based on the inverse relationship between impedance and body volume,7 allowing the estimation of the total blood volume status of the body or of specific areas (mainly intrathoracic by placing the electrodes over the corresponding area).21
Intrathoracic BI has only been tested in HF patients with reduced LVEF, using sensors embedded in resynchronisation devices or implantable defibrillators22, and the results of 2 more recent clinical trials question its usefulness.23,24
Non-invasive methods to assess total body impedance include whole body BI analysis (BIA) and bioelectrical impedance vector analysis (BIVA).25 The value of BIA in the study of HF is limited because algorithms for estimating total body volume have been established from healthy populations, so when the degree of water retention in different tissues is variable, as in HF, this results in inaccurate estimates for specific body compartments.25
BIVA detects peripheral oedema before it becomes evident and can be useful for assessing fluid in patients with hypervolaemia.26,27 Its measurement does not need to be standardised in reference populations, like with BIA,26 and allows estimation of total and compartment-specific body composition.21,26
In HF, total body volume measurements using BIVA have been shown to be useful in increasing the accuracy of disease diagnosis.27,28
Only 3 studies have evaluated the prognostic role of these BIVA measurements in acute HF.13,29,30 In the first, volume removal guided by serial BIVA assessments during hospitalisation was associated with increased survival 90 days after discharge.13 The latter 2 are single-centre studies where BIVA values are independently associated with risk of readmission29 or death.29,30
Regarding leg volume measurements using BIVA, there is only one large study showing that in the general population, leg BIVA values are inversely associated with the risk of HF.11
Our study has some limitations. The main limitation is the size and heterogeneity of the sample studied, as it includes patients with the full clinical spectrum of HF, which may affect the power of the analysis. However, as in other studies with sample sizes similar to or smaller than ours,27,28 despite these factors we have been able to demonstrate the prognostic value of this marker, overall, and in all the subgroups evaluated, suggesting that this parameter has a high predictive value. On the other hand, the availability of natriuretic peptides during admission in patients in our cohort was limited and the use of the Framingham criteria may have limited diagnostic accuracy in patients with preserved LVEF. However, cardiac structural abnormalities (left ventricular hypertrophy or left atrial dilatation) or evidence of diastolic dysfunction was present in 35 (92%) of the 38 H F patients with LVEF > 50% and in the remaining 3 cases NT-pro-BNP values were >1,000 pg/mL. Finally, this is a single-centre study, which limits the external validity of the results obtained.
However, the follow-up of the cohort is extensive and the consistency of the results obtained with different methodological approaches (bivariate analysis, survival and multiple logistic regression) strengthens the conclusions drawn.
ConclusionsAnkle electrical bioimpedance values measured with the wireless, hand-held IVOL device may be an independent predictor of long-term mortality in patients hospitalised for acute HF. This prognostic value is maintained in HF with non-preserved LVEF and in patients with RV dysfunction.
The simple, non-invasive measurement of BI with the proposed device could in the future allow continuous outpatient monitoring of the progress of HF patients and help to identify decompensation at a very early stage, aspects that are currently being evaluated in a clinical trial, the preliminary results of which have recently been published.31,32
Ethical aspectsThe study protocol was approved by the Research Ethics Committee of the Virgen Macarena and Virgen del Rocío Hospitals in Seville. The data collected were anonymised according to the recommendations of the Data Protection Law. Patients signed the informed consent form.
Authors' contributionE. Gutiérrez-Carretero, A.M. Campos and A. Peña were responsible for patient enrolment and data collection. J. Rossel designed and performed the measurements with the IVOL team. F. Medrano and J.M. Praena carried out the statistical analysis. F.J. Medrano, E. Gutierrez and A. Ordoñez drafted the manuscript. All authors participated in the conception and design of the study, interpretation of the data and review and approval of the manuscript.
FundingThis study has been funded by the Instituto de Salud Carlos III (grants DTS19/00137 and DTS19/00134).
Conflict of interestAll authors declare that they have no conflicts of interest in relation to the content of this article.
