Materials and methods
Study design and setting
This was a prospective observational study conducted in a tertiary care emergency department. The institutional review board approved this study (Protocol # 14172). Informed consent was obtained from patients before enrolling them in the study. End-stage renal disease (ESRD) patients presenting to the emergency department requiring emergent hemodialysis were enrolled. Emergent hemodialysis was carried out based on criteria of fluid overload and hypoxia (oxygen saturation < 89%), severe acidosis (serum bicarbonate < 10 mmol/L) or severe hyperkalemia (potassium > 6 mmol/L). Patients were recruited from July to October 2017. Inclusion criteria were an age of 18 years or older and completion of a full hemodialysis session with removal of at least 2 L of fluid. Exclusion criteria included patients who were placed on non-invasive positive pressure ventilation (NIPPV), patients who were unable to tolerate the passive leg raise maneuver, patients with arrhythmias, and patients with evidence of aortic regurgitation and mitral stenosis.
Outcomes
The primary outcome was the sensitivity and specificity of mitral valve VTI in detecting fluid responsiveness. Patients who had > 12% increase in their LVOT VTI after the passive leg raise were labeled as volume responders. Secondary outcomes included difference in the delta MV VTI pre- and post-dialysis. The delta VTI was defined as the percent change in VTI after the passive leg raise maneuver. As well as the differences in the delta LVOT VTI pre- and post-dialysis, the sensitivity, specificity, positive and negative predictive values of the MV VTI in detecting volume responsiveness. A subgroup analysis was done according to fluid removal during dialysis to look for mean delta VTI per fluid removal.
Interventions
Study investigators were two emergency physicians with fellowship training in emergency ultrasonography and an Emergency Ultrasound fellow. The study was done at an institution with an emergency ultrasound fellowship program with special emphasis on cardiac echocardiography. All investigators completed ten ultrasound scans with the same measurements done on each scan. All three investigators reviewed all scans to standardize approach and technique before enrolling patients. All imaging measurements were performed with patients seated on a stretcher with their legs parallel to the ground and the head of the bed elevated at 45°. Before the hemodialysis session, an apical four-chamber view was obtained and a Doppler tracing of the MV blood flow was recorded with the Doppler sampling gate placed proximal to mitral valve leaflets. Following that, an apical five-chamber view was obtained and a Doppler tracing of the LVOT blood flow was recorded with the pulse wave (PW) Doppler sampling gate placed proximal to the annulus. The velocity time integral was calculated using the average of five repeated measurements over one respiratory cycle. Ultrasonographic images were obtained with a 1–5 MHz phased array transducer on a SonoSite Xporte (SonoSite, Bothell, WA).
Measurement
Study participants’ age, sex, blood pressure, pulse rate and oxygen saturation pre- and post-dialysis as well as before and after passive leg raise (PLR) were recorded. History of co-morbidities was obtained from their medical records. Patients’ MV and LVOT VTI were measured before hemodialysis. After these measurements, a passive leg raise maneuver was performed. The head of the bed was lowered to the flat position and the patients’ legs were elevated 45° above the level of the heart. After 1 min, another set of Doppler tracing through the MV and LVOT was obtained, and the parameters were measured again. Delta VTI was defined as (VTI PLR − VTI initial)/VTI PLR. Another set of vital signs was taken after the passive leg raise maneuver. Patients were subsequently placed in the neutral position and taken to hemodialysis. After hemodialysis, a new set of vital signs, Doppler tracings and measurements were obtained before and after passive leg raise. All images were saved as still pictures, with measurements. The velocity time integral was obtained with electronic calipers in the ultrasonographic machine’s software by tracing the Doppler flow signal.
Analysis
Based on the literature, a patient is labeled as a volume responder if their LVOT VTI increases by 12% following the PLR [
14,
18‐
22]. Given that this is a pilot study looking at mitral valve VTI, patients with an increase in their MV VTI of ≥ 12% and < 12% were classified as responders and non-responders. Assuming a random sampling of patients, a power of 0.8 and an
α of 0.05, a minimum sample size of 43 cases was needed. All data were analyzed using SPSS for Windows, version 17.0 (SPSS, Inc, Chicago, IL). After meeting the normality assumption (Shapiro–Wilk test), differences between values before and after hemodialysis or between supine and passive leg raise positions were calculated. Quantitative variables are presented as mean ± SD plus 95% confidence interval (95% CI) and categorical variables as frequency (percentage). Paired
t test was performed to assess the statistical significances observed in VTI, HR, and MAP before and after hemodialysis. Independent
t test was used to assess the statistical significance of differences observed between groups of patients with different amounts of fluid removal.
Discussion
The results of this study have shown that in hemodialysis patients, non-invasive bedside echocardiography coupled with a passive leg raise is helpful in predicting volume responsiveness. In the pre-dialysis cohort of patients, the passive leg raise maneuver did not result in any VTI changes, whether at the level of the LVOT or at the mitral valve, whereas, after the fluid was removed, the delta LVOT VTI increased by 15% with PLR and the delta MV VTI increased by 12.95%. Our results also showed that the delta VTI increase was volume dependent as patients who had greater than 4 L removed had the greatest increase in delta VTI.
Fluid responsiveness and fluid management in shock are important topics for intensivists and emergency physicians. Aggressive hydration in the ED became the mainstay of septic shock therapy after the EGDT trial [
1]. The Rivers protocol was recently criticized and challenged by several studies that showed that aggressive fluid therapy in sepsis can be dangerous and can lead to increased mortality [
2‐
4,
23,
24]. Furthermore, the use of static measures such as the central venous pressure (CVP) in predicting fluid responsiveness has been put to question [
5,
6]. Fluid responsiveness is an even more pivotal issue in patients who are chronically fluid overloaded such as ESRD patients. When these patients present hypotensive due to sepsis, it is difficult for the treating physician to gauge their intravascular volume status through the use of physical exam findings and through static measures (mean arterial pressure, CVP) [
25,
26]. This has prompted researchers to look for new methods of assessment of volume responsive. One of the first methods looked at aortic flow changes using an esophageal Doppler. While it showed a direct correlation with intravascular volume changes, this technique remained highly invasive, as the patient needed to be intubated or sedated [
12]. More recently, there has been a gradual shift towards non-invasive dynamic ways of assessing volume responsiveness. Physicians looked at the inferior vena cava (IVC) collapsibility as a surrogate marker of volume responsiveness. While early studies showed great correlation between IVC collapsibility and patient’s volume status, the majority of studies did not show a great correlation with fluid responsiveness in spontaneously breathing patients [
8‐
10,
27‐
29]. More recently, a group lead by Lamia et al. [
13] looked at the role of non-invasive echocardiography as a measure of volume responsiveness in ICU, and they showed that a 12.5% increase in VTI was 77% sensitive and 100% specific for detection of a > 15% in cardiac output following volume expansion. These similar findings were reproduced by Maizel et al. [
14] who showed that in spontaneously breathing patients presenting in shock, they showed that a 12% increase in stroke volume after passive leg raise was 69% sensitive and 89% specific for response to 500 mL of crystalloid administration. To the best of our knowledge, our study is the first study looking at the role of non-invasive cardiac echocardiography in the emergency department for the evaluation of volume responsiveness. A study by Dinh et al. in 2012 showed that emergency physicians can accurately measure LVOT VTI and cardiac output [
30]. Our study remains the only study that looked at the role of MV VTI as a predictor of volume responsiveness. MV VTI was found to be highly specific for fluid responsiveness as well as having a high positive predictive value for detecting a volume-responsive state. This technique can be an alternative for physicians to evaluate volume responsiveness in cases where patients’ body habitus prevents them from getting an adequate apical five-chamber view. It is important to note that there were four patients that had a MV VTI < 12% while having an LVOT VTI > 12%. Possible explanations for the discrepancy could be that the patients had diastolic dysfunction or mitral regurgitation, two conditions that could affect mitral valve VTI. We did not, however, check for diastolic dysfunction on our patients. Furthermore, one patient had a MV > 12% and an LVOT VTI < 12%. This particular patient had an aortic valve replacement and his low LVOT VTI could be due to the metallic valve.
In contrast to the prior VTI studies, we removed fluid via hemodialysis and we looked at the effect of the passive leg raise once these patients had this fluid taken off. The rationale was that in the pre-dialysis cohort, patients would be hypervolemic and their overstretched heart will not respond to the PLR maneuver and the preload bolus, in contrast to the post-hemodialysis cohort, which would represent a fluid responsive state and would allow the heart to respond to the preload challenge and increase its stroke volume. The passive leg raise maneuver by definition is a reversible auto-bolus of about 200–400 mL. It has been studied in both ventilated and spontaneously breathing patients, and several studies have shown that an increase in stroke volume of 15% following a PLR maneuver had a specificity of 93% in detecting volume responsiveness after receiving a 500-cc fluid bolus [
21,
22,
31]. Furthermore, we chose to conduct this study on end-stage renal disease because this subset of patients represents a challenge while being evaluated for fluid responsiveness, as these patients are chronically fluid overloaded and still can present with hypotension and it is often unclear whether these patients require more fluid or vasopressor therapy.
It is interesting to note that there were no differences in vital signs before and after hemodialysis except for systolic blood pressure, which we noted to be higher by 8 mmHg in the post-dialysis cohort. This increase in systolic blood pressure after dialysis was described by Inrig et al. [
32] who explained that intradialytic hypertension is multifactorial, and its causes include subclinical volume overload, sympathetic overactivity, activation of the renin angiotensin system, endothelial cell dysfunction, and specific dialytic techniques. It is important to note that this increase in blood pressure is only temporary. In the majority of patients, there is a drop in afterload post-dialysis coupled with an increase in their cardiac output [
33]. This could explain why our VTI values increased in the post-hemodialysis cohort. However, it is more important to note that there were no changes in vital signs after the passive leg raise signs which further strengthens the idea that vital sign changes are not adequate enough for assessing volume responsiveness. This has been shown several times in the literature and in previous studies on bedside echocardiography and further strengthens the argument that VTI changes are more sensitive during volume changes than vital sign changes [
12,
13,
22].
The greatest increase in the delta VTI for both the mitral valve and the LVOT in our population was seen in patients who had more than 4 L of fluid removed. This can probably be due to several reasons; first, the interval between dialysis sessions was longer in our patients (7.8 days) than regular hemodialysis patients. Our patients receive hemodialysis on a compassionate or emergent basis based on criteria of fluid overload and hypoxia, severe acidosis (serum bicarbonate < 10 mmol/L0 or severe hyperkalemia (Potassium > 6 mmol/L). As such, they usually go more than the standard 3 days without dialysis and might be more fluid overloaded than the scheduled hemodialysis patient. Another possible explanation is that hemodialysis is done over a period of 3–4 h which could allow the system time to slowly adapt to the volume loss.
The present study is an observational study performed on a specific group of ESRD patients with volume overload. Therefore, its findings should be interpreted cautiously as they are not applicable to the general ED population. Our results should be compared with the results of future studies on other populations with hypervolemia or hypovolemia. Furthermore, our study is limited by its small sample size; however, this sample size was calculated based on the existing literature on volume responsiveness and velocity time integral. Moreover, considering ESRD patients after hemodialysis as euvolemic may not be accurate. Our study is also limited by the lack of lung ultrasonography. Although our aim was to investigate the value of mitral valve VTI, several studies have looked at the role of lung ultrasound in ESRD patients as a surrogate for volume status [
34‐
36]. Future studies need to combine cardiac echo with lung ultrasonography in an effort to best understand volume responsiveness. Finally, the ultrasound examination was performed by physicians with extensive training in bedside ultrasound and therefore cannot be generalized to all emergency physicians.
Authors’ contributions
RBC, JW and JC have made substantial contributions to conception and design of the study. RBC, JW, JC, GAD and SK: acquisition of data, analysis and interpretation of data. RBC, GAD and SK have been involved in drafting the manuscript. RBC, JW and JC: revising manuscript critically for important intellectual content. All authors contributed substantially to its revision. GAD and RBC take responsibility for the paper as a whole. All authors read and approved the final manuscript.