We conducted a cross-sectional study including 76 patients (68.4% male) from the dialysis unit of a secondary hospital with 103 hemodialysis patients. Patients that were more than two months in the technique in stable condition and without hospital admissions during the previous two months were included. Patients with contraindication for BIS (implanted electronic device, metallic prostheses of any type, amputated patients, pregnant or lactating women) were excluded. Six patients (7.3%) with a previously known severe valvulopathy or bad acoustic window also were excluded. Patients were 18 years old or above and had signed an informed consent approved by the Institutional Ethics Committee of Severo Ochoa Hospital.

Blood samples for standard laboratory parameters and BIS were obtained in the same day, before the second dialysis treatment of the week and after 20min in semi-recumbent position. Patients underwent echocardiography on the day following midweek dialysis.

Demographic and clinical characteristics were recorded. Dry weight was determined by the patient's attending physician in each hemodialysis session without knowledge of the bioimpedance results. Under these conditions bioimpedance was performed and weekly TAFO measured. Pre- and postdialysis body weight and blood pressure were collected. Intradialytic symptoms were noted and classified as hypotension, cramps, or other. Hypertension was defined as blood pressure greater than 140–90 and or antihypertensive therapy. Interdialytic weight gain and ultrafiltration rates in mL/h/kg were recorded.

Bioimpedance was assessed using a Body Composition Monitor (Fresenius Medical Care, Deutschland GmbH). Patients were measured before dialysis in a semirecumbent position in their dialysis chair. Following measures were taken: Fluid overload (FO), extracellular water (ECW), intracellular water (ICW) in liters, ECW/ICW ratio, lean tissue mass in percentage (LTM %), lean tissue index (LTI) defined as the quotient of LTM/height2 (kg/m2), fat mass in percentage (FAT %), fat tissue index (FTI) defined as the quotient of FAT/height2 (kg/m2) and phase angle (PA) in grades. Phase angle is calculated as the arctangent of reactance over resistance at frequency 50kHz and is related to body cell mass and to the distribution of fluid between intracellular and extracellular compartments.11

Relative fluid overload (FO/ECW) and TAFO were calculated. For calculating TAFO, FO was additionally estimated for each of the other two treatment days of the week by using differences in predialysis body weight (preweight) according to next equation:

FO BIS day indicates real measurements. This approach assumes that treatment-to-treatment changes in preweight over a few days reflect mainly changes in FO and that body composition remains stable within this short time period. Postdialysis FO was calculated by subtracting ultrafiltration volume from the predialysis measurement.

Fluid status in patients on intermittent ultrafiltration therapy is time-dependent, being lowest just after dialysis and reaching a maximum before the next dialysis session. In addition, fluid status is usually higher after the long interdialytic interval. To make different measurements of fluid status during the week comparable, the concept of weekly TAFO was introduced. It represents the average cardiovascular fluid load over one complete week, assuming linear fluid accumulation in the interdialytic period. It is defined as the average between all three FO predialysis and postdialysis FO (FOpost) values of the week.

Average weekly TAFO: (FO pre1+FO pre2+FO pre3+FOpost1+FOpost2+FOpost3)/6.12

In other studies, different TAFO values have been used to determined hydration status among patients.8,12 In this study, patients were divided into 3 groups based on time-averaged fluid overload: Group A (dehydrated), with TAFO <−0.25L; Group B (normohydrated), with TAFO between −0.25 and 1.5L; Group C (overhydrated), with TAFO >1.5L.

2D, M-mode and Doppler echocardiography was performed on patients lying in the left decubitus position at baseline, on the day after dialysis, by an experienced group of sonographers from the echocardiography laboratory, using a General Electric Vivid7 echocardiographic system with a 5S transthoracic probe. Registries were reviewed by a single sonographer, and all of them were blinded to the clinical details of patients. All echocardiographic data were measured according to the guidelines of the European and American Societies of Echocardiography.13,14 Left ventricle (LV) mass was calculated using Devereux formula and indexing by body surface area. Left atrial dilation was considered for a LAV >34ml/m2. Diastolic function measurements were assessed14 from the apical view using transmitral power-Doppler. Right ventricle diameter and function15 were determined from apical view measuring subvalvular diameter, tricuspid annular plane systolic excursion (TAPSE) and tricuspid annulus systolic velocity by tissue-Doppler.

Data were processed with IBM SPSS19 program. Quantitative data are expressed by mean (median for time variables) and a confidence interval of 95%, and qualitative variables by the number of subjects (n) and its relative percentage. Groups were defined from time-averaged fluid overload (TAFO) and the OH/ECW ratio, and were compared by one-way ANOVA and linear multivariate regression analysis (quantitave variables), and Chi-square for trend (qualitative variables). A variation >10% was settled for confounding factors. Correlations were compared by Spearman's rho. Statistically significance was considered when p<0.05.

Among the 76 patients included the median age was 63.1 (CI 95% 60.0–66.2) years, men 52 P (68.4%) with a median dialysis vintage of 36 (13.5–84) months. 31 patients (40, 8%) were diabetics, 23 patients (30.3%) were hypertensive. The underlying renal disease was Diabetes 21 P; Vascular disease 17P; Cystic hereditary disease 3P; Glomerulopathy 14P; interstitial 7 P; Unknown 12 P; Others 2 P. Regarding dialysis technique, 55 P (72.4%) patients were in conventional hemodialysis and 21 (27.6%) patients were in on-line hemodiafiltration. The vascular access used was native arteriovenous fistula in 40 (52%) patients, Graft in 8 (10, 5%) patients and permanent central catheter in 28 (37%) patients.

Based on time-averaged fluid overload 32 (42%) patients were considered dehydrated; 26 (34%) patients were considered normohydrated and 18 (24%) patients were considered overhydrated. Table 1 summarizes general characteristics according TAFO groups. We found no differences between groups regarding age, systolic and diastolic blood pressure, diabetic history and dialysis vintage; regarding sex, men more often out of range (dehydrated or overhydrated). Additionally, no differences were found with the dialysis technique, the frequency of intradialysis hypotension and ultrafiltration rates. Interestingly, we founded that interdialytic weight gain was higher in dehydrated and overhydrated patients compared to normohydrated patients. As expected, in dehydrated group, ECW/ICW index was lower and phase angle was higher. While there was no difference of lean mass and fat mass percentage, LTI was higher in dehydrated patients. In regard of biochemical parameters, there was no difference between groups in serum sodium and creatinine; however, as usual, albumin and hemoglobin levels were lower in overhydrated patients.

Table 2 shows data about echocardiographic parameters. Left atrial dilatation (LAVI >34ml/m2) was present in 24 (33.3%) patients without atrial fibrillation, being moderate in 3 (4.2%) and severe in 8 (18.1%). We found a significant correlation between the hydration status measured by TAFO and left atrial volume index (LAVI) (Spearman's rho 0.29; p=0.013). Interestingly, this correlation was not found when the hydration status was defined by FO/ECW (Spearman's rho 0.06; p=0.61), in spite of the significant internal correlation between both measurements (TAFO and FO/ECW) (Spearman's rho 0, 46; p<0.0001). Three patients with atrial fibrillation (3.9%, one from Group A and 2 from Group C) were excluded of the analysis.

Using One-Way ANOVA test and linear regression methods, the TAFO defined hydration status, but not the FO/ECW, kept a significant relationship with LAVI (p=0.03). When including age, sex, ethiology, dialysis vintage, comorbidities, dialysis technique, hypertension, biochemical parameters (Hemoglobin, Na, K, Albumin, Creatinin), and all echocardiographic measurements including valvular regurgitation and left ventricle mass index, the only confounding factors identified were lean and fat mass percentages related to total weight, and diastolic function measured by transmitral velocities ratio, but the adjusted relationship kept significant (p=0.036).

Plasma sodium was not different between the 37 patients with (136.6mEq/l; CI 95% 135.8–137.4) and 35 patients without (137.3mEq/l; IC95% 136.1–138.6) atrial dilatation (p=0.314).

Left ventricle hypertrophy was present in 73.7% of patients and concentric remodeling in 7.9%. Concentric hypertrophy was present in 63.2% of patients and eccentric hypertrophy only in 10.5%. No statistical difference between TAFO defined groups (p=0.838) nor regarding left ventricle mass index (p=0.29) was found. Left ventricle ejection fraction (LVEF) was normal (in fact, all groups tended to hypercontractility), but in two cases from group A (LVEF 36.47 and 45.2%, moderate and mild systolic dysfunction respectively) and one from group C (LVEF 46.7%, mild systolic dysfunction). 48 patients (66.7%) had a pattern of diastolic dysfunction type I, defined by tissue-Doppler study of mitral annulus and power Doppler transmitral velocities, with no difference between groups (p=0.401). There was no significant difference in the prevalence of interdialysis hypertension (Group A 21.9%, Group B: 42.3%, Group C: 27.8% (p=0.241)).

No severe valvulopathy was found: Moderate mitral regurgitation was present in 9.2% of patients and moderate aortic regurgitation in 2.6%, without differences between groups. Mild pericardial effusion was present in only 5.3% with no significant difference between groups. No pulmonary hypertension was present, although pulmonary systolic pressure could only be determined in 40.8% of patients.