Authors: Melissa C. Caughey, Dhatri Kakarla, Sora Lee, Simran K. Singh, Caroline J. Poulton, Emily H. Chang
Categories: Research, End-stage kidney disease, Dialysis, Volume status, Noninvasive monitoring
Source: BMC Nephrology
Authors: Melissa C. Caughey, Dhatri Kakarla, Sora Lee, Simran K. Singh, Caroline J. Poulton, Emily H. Chang
Fluid volume overload is a frequent cause of hospitalization for patients with end-stage kidney disease (ESKD), and disproportionately occurs during the 2-day hemodialysis gap when patients are most susceptible to poor volume control. Noninvasive, wearable devices may hold potential for at-home volume status monitoring to prevent hospitalization but clinical validation in ESKD populations is limited.
A total of 17 study participants (41% female, 76% Black, median age = 68 years) were recruited for in-center testing of hydration status pre- and post-hemodialysis. Percent lung water (%LW) and skin hydration were noninvasively assessed and compared with established clinical and hemodynamic and markers of hydration status.
Median within-subject change in systolic blood pressure (-10 mmHg, p = 0.08), diastolic blood pressure (-6 mmHg, p = 0.06), weight (-1.6 kg, p < 0.0001), heart rate (-10 bpm, p = 0.09), and %LW (-1.0%, p = 0.03) declined from pre- to post-dialysis. When analyzed as a continuous measurement, %LW was not correlated with systolic blood pressure, weight, heart rate, or dyspnea score. Skin hydration demonstrated poor correlation with %LW, but statistically differed (42 vs. 23 a.u.; P = 0.03) when stratified by the broader categories of borderline-high/elevated versus normal %LW, suggesting an ability to detect abnormal fluid status.
In this small pilot study, noninvasive assessment of %LW was sufficiently sensitive to detect pre-to-post hemodialysis change in fluid status. These results support future evaluations in larger studies of patients with ESKD. Noninvasive, wearable devices may be a feasible strategy to assess volume status in the ESKD population, an approach which could potentially be extended to at-home settings.
The online version contains supplementary material available at 10.1186/s12882-025-04306-w.
The incidence of end-stage kidney disease (ESKD) in the United States is projected to increase by 11–18% in the next 10 years, and by 2030 the number of Americans living with ESKD is expected to exceed 1.2 million [1]. On average, patients with ESKD who are managed by hemodialysis (HD) experience 1.7 hospitalizations per year [2]. Fluid volume overload is a frequent cause of hospitalization, with associated Medicare costs exceeding $100 million per year [3, 4]. Admissions to intensive care units for volume overload disproportionately occur during the 2-day HD gap (on Sunday, for patients dialyzed on a Monday, Wednesday, Friday [MWF] schedule, and on Monday for those dialyzed on the Tuesday, Thursday, Saturday [TTS] schedule), when patients are most susceptible to poor volume control [4]. The heightened risk of hospitalization suggests a potential need for at-home fluid status monitoring during the 2-day HD gap.
At-home volume monitoring accompanied by an as-needed, interdialytic telehealth intervention, such as physician advisement to restrict fluids, increase diuretic dose (for patients with residual kidney function), or referral to outpatient HD (in the event of severe interdialytic volume overload), may optimize volume control in patients with ESKD and potentially reduce hospitalizations for acute pulmonary edema and volume overload [5]. Noninvasive, at-home, wearable monitors, such as the Remote Dielectric Sensing (ReDS) vest (Sensible Medical Innovations, Inc.; Raleigh, NC) have been associated with reduced hospitalizations for patients with heart failure, but their role in volume status monitoring for patients with ESKD is less well established. Wearable, skin hydration sensors may also hold promise for at-home volume status monitoring [6, 7], which may be applicable to ESKD [8]. The skin is an easily accessible organ, irrespective of body habitus, and a reliable source of biomarkers indicative of systemic conditions [9]. The purpose of this study was to evaluate the sensitivity of 2 noninvasive, commercially available instruments, for detection of fluid volume change in patients with ESKD, when validated against fluid removal induced by hemodialysis. To examine whether noninvasive and easily obtained assessments of volume status correlate with established markers of hydration in patients with ESKD, we monitored volume status using a ReDS vest and a MoistureMeter epidermal hydration sensor (Delfin Technologies; Miami, FL) before and after HD. The ReDS vest and the MoistureMeter have not been clinically validated in ESKD populations.
Patients with ESKD who were managed by conventional hemodialysis were recruited with informed consent from a dialysis center affiliated with the University of North Carolina. We excluded any of the following conditions which confound reliable measurements of lung water by a wearable Remote Dielectric Sensing (ReDS) height < 155 cm or > 190 cm, body mass index < 22 kg/m^2^ or > 38 kg/m^2^, active pneumonia, lung carcinoma, history of pulmonary embolism, pacemaker on the right side of the chest, and physical deformity, injury, or recent surgery to the thorax area that may prevent proper vest application or adjustment [10, 11]. Due to the height and body habitus requirements for the ReDS system, our study population was limited to adult patients ≥ 18 years. We also excluded any patients who were pregnant, or unable to provide informed consent. All clinical data were collected in the dialysis center before and after HD, with institutional review board approval.
Demographics, clinical data, hemodialysis measurements, and medications were abstracted from the dialysis center records. Conventional indices of hydration and volume status or related symptoms included target weight and successful attainment of target weight, as well as hemodynamic parameters (e.g., systolic and diastolic blood pressure, heart rate). Patient reported symptoms of perceived dyspnea were ascertained using the Dyspnea Visual Analog Scale (VAS). The dyspnea VAS is a validated instrument that is suitable even in the event of low literacy levels [12], and ranges from 1 (“not breathless”) to 10 (“extremely breathless”). All markers of volume status or associated symptoms were measured directly before and after hemodialysis.
Percent lung water was quantified using the Remote Dielectric Sensing (ReDS) system. The ReDS vest is a noninvasive, wearable device capable of detecting clinical and subclinical pulmonary congestion. Percent lung water (%LW) measured by the ReDS system has been validated against computed tomograph of the lung, with an intraclass correlation coefficient of 0.95 in preclinical studies [13] and 0.90 in evaluations of patients with heart failure [14]. The ReDS system has also been validated against invasive assessment of hemodynamics in patients with heart failure, demonstrating good diagnostic accuracy for the detection of elevated pulmonary capillary wedge pressure, a clinical surrogate of volume overload [11]. Percent lung water was recorded before and after HD by trained personnel, following the manufacturer’s instructions.
Skin hydration was quantified using a commercially available, hand-held bioelectric impedance device (MoistureMeterD) [15]. With this device, the tissue dielectric constant, corresponding to the amount of free and bound water, is measured by transmitting an electromagnetic wave and quantifying the electrical impedance between two cutaneous electrodes. The MoistureMeterD system includes multiple open-ended coaxial probes, which differ by the spacing between the inner and outer electrodes, allowing various skin measurement depths [15]. Because the epidermis is avascular, hydration was evaluated from the dermis, before and after HD, using the “EpiD” probe. The EpiD probe has an effective measurement depth of 0.5 mm, corresponding to the dermis, and does not include subcutaneous edema. Measurements were made in triplicate at the inner wrist, to evaluate measurement repeatability. The arm without vascular access was used, with skin markers to designate the measurement site for repeated pre- and post-dialysis testing. The MoistureMeter contains a pressure sensor which indicates if the device is held too firmly against the skin, to standardize the data collection.
All statistical analyses were carried out using SAS version 9.4 (SAS Institute; Cary, NC). Non-parametric tests were used for the statistical analysis, due to the small sample size of study participants. Within-subject differences in clinical measurements and markers of volume status, before and after hemodialysis, were analyzed using Wilcoxon sign-rank tests. Spearman correlations of %LW with conventional measures of volume status and skin hydration were analyzed, both at pre-dialysis and post-dialysis, as was the difference in pre-to-post %LW with pre-to-post measures of volume status and skin hydration. Distributions of skin hydration measurements, stratified by borderline-high / elevated %LW versus normal %LW, were compared using Wilcoxon rank sums testing. Because of the small sample size of this pilot study and limited statistical power, a P-value < 0.10 was considered suggestive of a statistically meaningful association.
A total of 20 patients were enrolled and completed volume status testing before and after hemodialysis. Of these, 2 failed to meet the BMI requirement for the ReDS vest and were excluded due to likely confounding or inaccuracy of the %LW measurement. A third study participant refused %LW testing due to an in-dwelling catheter placed on the right chest and apprehension about wearing the vest. After excluding these 3 study participants, our final study population consisted of 17 patients with a median age of 68 years; 41% of which were female, and the majority of which were Black (76%). Clinical diagnosis of hypertension was highly prevalent (94%), as was diabetes mellitus (53%), history of coronary artery disease (18%) and heart failure (18%), Table 1. Most study participants were receiving statins (65%), calcium channel blockers (59%), or beta blockers (53%), with 12% receiving angiotensin converting enzyme II inhibitors and 18% receiving aldosterone receptor blockers. 18% were receiving diuretics. A comparable proportion of study participants presenting on MWF versus TTS hemodialysis schedules (47% vs. 53%). Approximately a third (35%) completed volume status testing the day after the 2-day hemodialysis gap (i.e., on Monday or Tuesday, respectively, for the MWF and TTS schedules), when volume management is expected to be worst.
Table 1Demographic and clinical characteristics of the study populationCharacteristicNumber (%) orMedian (25th – 75th percentile)N = 17 Demographics Age (years)68 (52–74)Sex (female)7 (41%)Race (Black)13 (76%) Medical History Hypertension16 (94%)Diabetes mellitus9 (53%)Liver disease1 (6%)Heart failure3 (18%)Coronary artery disease3 (18%)Cancer2 (12%)Stroke0 Hemodialysis Target weight (kg)78 (65–84)Prescribed treatment time (minutes)225 (210–240)Administered treatment time (minutes)217 (208–230)Ultrafiltration goal (mL)1750 (800–2250)Ultrafiltration rate (mL/min)7.9 (3.6–9.8) Medications Angiotensin II converting enzyme inhibitor2 (12%)Aldosterone receptor blocker3 (18%)Diuretic3 (18%)Beta blocker9 (53%)Calcium channel blocker10 (59%)Vasodilator2 (12%)Alpha blocker1 (6%)Insulin5 (29%)Statin11 (65%)
Overall, both study devices (ReDS vest and MoistureMeter) were well tolerated, with no adverse events or unanticipated problems. Perceived dyspnea scores were low at initial presentation (median VAS = 1.0), suggesting an absence of symptomatic hypervolemia; however, 5 participants (29%) had borderline-high or elevated %LW prior to dialysis, as measured by the ReDS monitor. From pre- to post-dialysis, median within-subject decreases in systolic blood pressure (-10 mmHg), diastolic blood pressure (-6 mmHg), weight (-1.6 kg), and heart rate (-10 bpm) were observed (Table 2). Along with conventional markers of volume status, the median within-subject %LW (-1.0%) and skin hydration (-1.3 a.u.) also declined from pre-to-post dialysis, Table 2. The majority of participants achieved the prescribed hemodialysis treatment time, with no significant difference in prescribed vs. delivered treatment. However, 6 participants (35%) had post-dialysis weights exceeding the target weight. The median excess weight above the target weight was 0.7 kg and ranged from 0.3 to 1.5 kg. When stratified by presentation day (3-day gap vs. 2-day gap), pre-dialysis weight above target weight was marginally higher for participants evaluated during the 3-day gap (2.1 vs. 1.2 kg, P = 0.1), but group differences in pre-dialysis %LW (32 vs. 31%, P = 0.9) and skin hydration (28 vs. 26 a.u., P = 0.7) were not significant, Supplemental Table S1.
Table 2Clinical, hemodynamic, and measured biomarkers of volume status, pre- and post-hemodialysisCharacteristicPre-DialysisMedianPost-DialysisMedianWithin-SubjectChange (Δ)P-value*Systolic blood pressure (mmHg)152 (125–171)152 (117–159)-10 (-21 to + 3)0.08Diastolic blood pressure (mmHg)87 (73–101)66 (58–88)-6 (-21 to + 3)0.06Weight (kg)78 (65–84)77 (64–84)-1.6 (-2.1 to -0.9)< 0.0001Heart rate (bpm)82 (73–89)74 (65–80)-10 (-15 to -6)0.09Dyspnea Visual Analog Scale1.0 (1.0–1.0)1.0 (1.0–1.0)0 (0 to 0)0.8Lung Water (%LW)31 (23–36)30 (25–32)-1.0 (-4.0 to 0)0.03Skin Hydration (a.u)27 (22–37)24 (21–37)-1.3 (-2.0 to + 0.7)0.4Treatment time (minutes)†225 (210–240)217 (208–230)0 (-17 to + 5)0.3Values expressed as medians (25th percentile – 75th percentile)*P-values for within-subject change, calculated by Wilcoxon signed rank tests†Treatment time at pre-dialysis = prescribed time, Treatment time at post-dialysis = actual delivered treatment time
When analyzed as a continuous measurement, %LW was not correlated with systolic blood pressure, weight, heart rate, dyspnea VAS, or skin hydration measurements, whether at pre-dialysis, post-dialysis, or when considering the pre-to-post dialysis changes in the measurements. A moderate and inverse correlation was observed between %LW and diastolic blood pressure at pre-dialysis, but a significant correlation was not observed at post-dialysis, nor when considering the correlation between pre-to-post dialysis changes in the measurements, Table 3. The pre-to-post dialysis change in %LW correlated with the change in relative percent blood volume (r = 0.77, P = 0.02), Supplemental Fig. S1; however, the analysis was limited to a subset of 9 study participants with available Crit-Line monitoring.
Table 3Correlations of percent lung water, as assessed by remote dielectric sensing (ReDS) vest, with hemodynamic, clinical, and measured biomarkers of volume status, pre- and post-hemodialysis, and aggregate measures across the Hemodialysis sessionCorrelate of %LWPre-dialysisPost-dialysisPre-to-post ΔCorrelation (r)Correlation (r)Correlation (r) Pre- and post-HD Systolic blood pressure-0.04 (P = 0.8)-0.10 (P = 0.7)-0.03 (P = 0.9) Diastolic blood pressure-0.52 (P = 0.03)-0.25 (P = 0.3)-0.11 (P = 0.6) Weight0.25 (P = 0.3)0.34 (P = 0.2)0.10 (P = 0.7) Weight above target0.28 (P = 0.3)0.17 (P = 0.5)0.10 (P = 0.7) Heart rate0.23 (P = 0.4)-0.01 (P = 0.9)-0.21 (P = 0.4) Dyspnea Visual Analog Scale0.05 (P = 0.8)0.05 (P = 0.8)-0.36 (P = 0.2) Skin hydration0.11 (P = 0.7)0.26 (P = 0.3)-0.12 (P = 0.7) Aggregate HD session Ultrafiltration volume-0.17 (P = 0.5) Dialysis treatment time0.16 (P = 0.5) Ultrafiltration rate-0.17 (P = 0.5)Abbreviations: %LW = percent lung water, r = Spearman correlation coefficient, HD = hemodialysisPredialysis percent lung water correlated with predialysis markers of volume status, postdialysis percent lung water correlated with postdialysis markers of volume status, and pre-to-post change in percent lung water correlated with pre-to-post change in markers of volume status or measurements aggregated across the hemodialysis session
Skin hydration demonstrated poor correlation with %LW when comparing same-subject measurements, both at pre-dialysis, post-dialysis, and change across dialysis (Table 3). However, when stratified by the broader categories of borderline-high/elevated versus normal %LW, distinct distributions of skin hydration measurements were observed, with statistically significant differences in median values (42 vs. 23 a.u.; P = 0.03), Fig. 1. The triplicate-measured MoistureMeter readings had high reproducibility, yielding a coefficient of variation of 5.6% at pre-dialysis, and 6.4% at post-dialysis.
Fig. 1Pre-hemodialysis distributions of skin hydration measurements, as assessed by the MoistureMeter device, stratified by borderline high or elevated (> 35) versus normal (≤ 35) percent lung water, as assessed by the Remote Dielectric Sensing (ReDS) vest
In this clinical validation study, we demonstrate that fluid status in ESKD populations can be evaluated by a noninvasive, wearable device, and in future work, hydration assessments could potentially be extended to at-home settings. The percent lung water, as assessed by the ReDS system, was sensitive to fluid volume change from pre- to post-HD. A general decline in mean within-subject skin hydration was also observed across pre- to post-HD. When grouped more broadly into categories of borderline/high versus normal %LW, skin hydration measurements statistically differed, suggesting the metric may have utility for detection of abnormal volume status. These preliminary findings support the concept that noninvasive wearable devices may have utility for at-home monitoring of patients who are at risk for interdialytic volume overload.
Various techniques are used to assess fluid overload in clinic-based settings. These include measurement of the inferior vena cava diameter by echocardiography, quantification of ultrasonic reverberation, or “B-lines” on lung ultrasound, assessment of total body water by segmental bioimpedance testing [16], and continuous relative plasma volume measurement during the HD session [17, 18]. However, patients managed by HD may also benefit by at-home monitoring of interdialytic volume, particularly during the 2-day HD gap when most susceptible to fluid volume overload. Existing at-home monitoring strategies, such as ambulatory blood pressure monitoring, may identify volume overload in some patients, but not all. In the setting of essential hypertension, blood pressure may be elevated despite normal fluid volume. Alternatively, fluid volume may be elevated in the setting of normal blood pressure readings, which could be masked due to reduced cardiac output or antihypertensive medication use.
Interdialytic weight gain (IDWG) can also be monitored at home by patient self-assessment, but may be an insufficient indicator of volume overload. When elevated, the IDWG might signify poor adherence with fluid restrictions [19]. However, IDWG is also influenced by residual post-dialysis volume status. Patients achieving post-dialysis volume depletion tend to have a greater IDWG than those with residual volume overload [20]. Monitoring IDWG alone also appears to have low prognostic utility by only predicting adverse events when extreme (> 4 kg) [21]. While at-home assessment of blood pressure or weight gain may not adequately identify interdialytic volume overload, continuous pulmonary artery pressure monitoring by sensors affixed to the interior pulmonary artery wall during right heart catheterization may be an option, and have demonstrated successful reduction in hospital readmissions for patients with heart failure [22]. The approach is invasive, however, and is not commonly implemented in patients with ESKD.
The prevalence of pulmonary congestion in patients with ESKD may be under-detected by patient-report. In our study, patient-reported perceived dyspnea prior to HD was minimal, but the detection of borderline/elevated %LW by the ReDS system approached 30%, suggesting the presence of subclinical volume overload. In previous work, we assessed interdialytic volume overload by echocardiography, in an observational registry of 778 patients with stage 4 or 5 chronic kidney disease at our institution [23]. Imaging was performed one day prior to scheduled hemodialysis, excluding the 2-day HD gap, for those managed by HD. Overall,19% had evidence of elevated left atrial pressure in the absence of mitral regurgitation and 15% were classified with moderate to severe left atrial dilation, which is suggestive of volume overload. The prevalence of pulmonary hypertension was 12%, with approximately half of pulmonary hypertension cases arising secondarily to elevated left atrial pressure [23]. When limiting the analysis to patients managed by HD, the prevalence of pulmonary hypertension was 16%. However, echocardiography is less feasible as a strategy for at-home pulmonary congestion monitoring.
If shown efficacious, at-home monitoring for pulmonary congestion by wearable devices could become an accepted and widely utilized telehealth strategy for the management of ESKD, because few changes to standard practice would be required for successful adoption. Patients with ESKD could maintain a conventional 3-day a week HD schedule, with at-home volume monitoring accompanied by an as-needed, interdialytic telehealth intervention, such as physician advisement to restrict fluids, increase diuretic dose (for patients with residual kidney function), or referral to outpatient HD (in the event of severe interdialytic volume overload), However, medical insurance coverage would be mandatory, both for the device used to monitor pulmonary congestion and for the billing of telehealth services. These costs could potentially be offset by reduction in hospitalizations for volume overload; however, a formal cost-effectiveness analysis should be conducted. Regardless, enhanced volume control and reduced hospitalizations would improve patient quality of life, and because patients could maintain a conventional HD schedule, the strategy is likely to have favorable patient acceptance.
However, the accessibility of at-home hydration status monitoring is an important consideration. The ReDS vest is limited by the BMI requirement, and although not considered a contraindication by the device manufacturer, in our pilot study, patients with an indwelling catheter were apprehensive about the device and declined study participation. This, and the BMI restriction, limit the applicability of the ReDS device to all patients with ESKD. Cutaneous devices such as the MoistureMeter are more broadly accessible, without the limitations by body habitus, but skin hydration assessment in patients with ESKD could potentially be influenced by skin salt stores. Sodium binds to glycoaminoglycans (GAGs) in the dermal interstitium, a process that is thought to be osmotically inactivating, because the sodium sequestration has no impact on body weight, blood pressure, or extracellular volume [24]. On the other hand, imaging studies using magnetic resonance sequences (23Na-MRI and H-MRI) before and after dialysis show that both sodium and water content of the skin decrease following dialysis [25]. In light of this, cutaneous assessment of dermal water content may plausibly reflect hydration status.
This pilot study evaluating noninvasive hydration monitors in patients with ESKD has some limitations, which merit consideration. Our study population consisted of well-managed and clinically stable study participants who consistently presented to their scheduled HD sessions, and did not include patients with a recent (< 6 months) transition to HD, or those with low adherence to HD, who would have greater vulnerability to volume instability and interdialytic hospitalization for volume overload. Due to limited study coordinator availability on Mondays, only one-third of the study population was evaluated during the 2-day HD gap, when patients are likely to be more volume overloaded. The pre-to-post difference in %LW (31 to 30%) was not clinically significant, but importantly, none of our patients presented with dyspnea or clinical symptoms of volume overload or signs of heart failure. In fact, the VAS dyspnea scale was the lowest possible score (median of “1”), both at pre-dialysis and post-dialysis. Although our demonstrated decrease of 1% in %LW is not clinically meaningful, it does show the sensitivity of the instrument to detect change in volume status, when validated against fluid removal by dialysis, even when no symptoms are present at pre-dialysis.
Along with the limitations inherent with analyzing a well-managed study population, our pilot investigation had other shortcomings as well. The time lag between ultrafiltration and capillary refill may have influenced hydration assessments immediately following HD; however, both study devices capture composite signals from the combined interstitial and vascular spaces, and are not localized to the vascular component. On the other hand, because this was small pilot study, our analysis of pre-to-post hydration status had limited statistical power for many of the comparisons. Despite these limitations, our study results suggest that noninvasive, wearable devices may be a feasible strategy to assess volume status in the ESKD population, an approach which could potentially be extended to at-home settings. In future work, the validity of volume status monitoring devices should be evaluated in a larger study of patients with ESKD. The feasibility of these devices in at-home settings should also be tested, and compared against other approaches, such as at-home IDWG monitoring.
Below is the link to the electronic supplementary material.
Supplementary Material 1