Diagnostic accuracy of internal jugular vein ultrasound in quantification of the central venous pressure for hemodialysis patients

Assessment of the central venous pressure (CVP) is an essential hemodynamic parameter for monitoring the dialyzing patients. Our objective of the present study is to investigate the accuracy of CVP measurement by internal jugular vein US in comparison to the direct measurement by the central venous catheters for hemodialysis patients. We included 106 patients; where their CVP was assessed in two different non invasive US methods (CVPni) separately and in combination and the obtained measurements were correlated to the invasive measurements (CVPi) by catheters. By method 1, there is a highly significant positive correlation between CVPni and CVPi (ρ < 0.001) and a Pearson correlation coefficient (r = 0.913 n = 93), and by method 2, there is also a highly significant positive correlation between the CVPni and CVPi in both groups (r = 0.832, 95%, n = 106, p < 0.001), 1.935 was the cut-off point for prediction of CVP ≥ 10cmH20. For differentiation between patients with CVP < 10cmH20 and ≥ 10cmH20, the accuracy measures (sensitivity, specificity, PPV, NPV, and overall accuracy) were 100%, 79.31%, 74.47%, 100%, and 87.10% by method 1, and were 91.11%, 85.48%, 82.00%, 92.98%, and 87.85% by method 2, while the combination of both methods had gained 88.57%, 89.66%, 83.78%, 92.86%, and 89.25%, respectively. The US offered a reliable and non-invasive tool for monitoring CVP. The present study has a novelty of combining more than one US method and this had reported higher accuracy measures and outperformed the use of a single method.


Background
Chronic kidney disease (CKD) is a worldwide health problem that has multiple etiologies. It could result in great morbidity and mortality thus exerting a high burden upon the health systems [1,2] Dialysis is the mainstay of treatment for renal failure patients and the prevalence of the dialysis-dependent population is expected to increase over time. However, the tolerability for fluid changes during the hemodialysis sessions is limited where the hypervolemia could result in pulmonary congestion, but on the other hand, fluid deficiency results in hypotension and could promote cardiac ischemia [3,4].
Continuous monitoring of the hydration status is considered of utmost importance and the definition of dry weight is a mandatory step in the patients' care [5]; assessment of the central venous pressure (CVP) is one of the used parameters to detect the hemodynamic changes and the patient hydration. As the mean CVP shows an initial significant decrease during the first hours of the dialysis session then attains a steady decline over the following hours, so the measurement of CVP in such patients can help to early detect the dry weight [6]. Measurement of CVP is usually done by a centrally inserted venous catheter, which is an invasive procedure that could carry multiple risks [7], so it is important to have a non-invasive and reliable method that could be easily used in daily practice from the start to the end of the dialysis session [8].

Open Access
Some authors had suggested that the CVP estimation by the ultrasound (US) examination of the internal jugular vein seems to be a promising surrogate for the direct invasive measurement, especially in the patients who have dialysis shunts and do not have available central venous lines [9,10].
The present study aims to investigate the accuracy of measurement of the CVP by internal jugular vein US for hemodialysis patients in comparison to its direct measurement by the already present central venous catheters in the same patients.

Subjects
The study was designed as a prospective cross-sectional analytical study, including 106 hemodialysis patients who were recruited from the nephrology and dialysis units in our institute, during the period from October 2020 to March 2021. The ultrasound studies were performed in the ultrasound unit of the radiology department in our institute.
• Inclusion criteria: patients above 18 years old who had a renal failure on hemodialysis with an already present central catheter. No catheter was inserted just for the study. • Exclusion criteria: patients with clinical conditions that cause increased right atrial pressure, and were not related to hypervolemia including the cardiopulmonary causes as well as the patients who had cervical or mediastinal masses or those with recent or old jugular venous thrombosis that was proved clinically or by imaging studies.
This prospective study was performed following the ethical guidelines of the Research Ethics Committee of our institute. The reference number: Code Ms-80-2020, Date of approval 16-08-2020; and it was approved by the local Research Ethics Committee of our institute. All of the participants were informed of the details and gave their written informed consent.
All of our patients were subjected to:

History taking
Every patient was requested to give the following data: • Cause of renal impairment.
• History of central venous obstruction by old catheter insertion or due to other causes. • Associated diabetes mellitus (DM), hypertension (HTN), or other medical disorders?

Non-invasive CVP measurement by high-resolution ultrasound
All patients were examined by ultrasound using a General Electric Logic P6 machine that is equipped with a linear probe with 7.5-12 MHz frequency.
The probe was applied gently on the skin with a copious amount of gel to avoid compression of the neck veins.
Patients were placed in a comfortable position in 45° reverse Trendelenburg orientation and their necks were slightly extended. The catheter-free IJV was used for taking measurements. In a longitudinal view, the average point of oscillation (collapse) during quiet normal spontaneous respiration was detected and marked on the skin, and then the height (vertical distance) between the collapsing point and the sternal angle was estimated.
The CVP was calculated by adding five cm (an average distance from the sternal angle to the center of the right atrial cavity) to the measured height of the IJV (at the marked collapsing point) then the results were obtained in cmH20 and recorded as CVPni.
Method 2 (Measurement of the cross-sectional area ratio between the internal jugular vein/common carotid artery) (Fig. 1b).
Our patients were seated in a flat supine position, then the largest cross-sectional area of IJV and the CCA were taken in their short axes at the level of the thyroid cartilage; the measurements were taken at the end of expiration to minimize the effect of the thoracic pressure. (Fig. 1c) We had performed the study on the patients who had already central venous catheters that were inserted for other purposes, but no catheters were inserted just for the study.

Invasive CVP measurement by an already present central venous catheter as the gold standard (CVPi)
A 3-way stopcock was connected to a fluid manometer through an intravenous fluid drip where the position of the H2O column was adjusted in such a way that the 3-way stopcock was at the same level as the right atrium (about 5 cm below the sternal angle). The intravenous fluid perfusion was directed in one of the stopcock ways to feed the column. The 3-way stopcock was then turned to make the intravenous column evacuating into the IJV catheter until the equilibrium was reached, and then the CVP was estimated at the level where the intravenous fluid had stopped in the column. The mean CVP value was taken during quiet normal respiration and was recorded in cmH2O.
For the statistical analysis, our study populations were divided into two groups based on a CVP cutoff point (that was 10 cm H2O), as a threshold for the prediction of (hydration) volume responsiveness (Figs. 2, 3).

Statistical methods and data analysis
Data were coded and entered using the statistical package for the Social Sciences (SPSS) version 26 (IBM Corp., Armonk, NY, USA). Data were summarized using the mean, standard deviation in quantitative data and using the frequency (count) and the relative frequency (percentage) for categorical data. Comparisons between the quantitative variables were done using the unpaired t-test. For comparing categorical data, Chi-square (χ2) test was done. Exact test was used instead when the expected frequency was < 5. Correlations between the quantitative variables were done using the Pearson correlation coefficient. ROC curve was constructed with the area under curve (AUC) analysis performed to detect the best cutoff value of ADC for detection of CVP ≥ 10 cmH2O (recommended threshold for predicting hydration) using the 2 methods. Standard diagnostic indices including sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and the diagnostic efficacy were calculated. Ρ-values < 0.05 were considered statistically significant. Logistic regression was done to detect the predicted probability of the combination of both methods (methods 1&2).

Results
One hundred and six patients were enrolled in this study; of which 35 patients were diabetic (33%), 35 patients were hypertensive (33%), 72 patients had their left side examined (67.9%), while 34 patients had the right side (32.1%). The mean CVPi was 9.55 ± 4.17 cmH20. Sixtytwo patients had a CVPi less than 10 cmH2O (58.5%) (Fig. 4) and forty-four patients had a CVPi more than or equal to 10 cmH2O (41.5%) (Fig. 5); However, there was no significant correlation between patients' gender, presence of DM, presence of HTN or the examined side, and the obtained CVPi (Table 1).

For method 1
It worth mentioning that the exact measurements of CVPni could not be obtained in nine of our patients, including those who were having a very high CVPi (their mean CVP was 19.278cmH20) as the IJV was distended all-through its neck course (Fig. 6), and for four patients who were having a low CVPi (their mean CVP was 2.25cmH20) (Fig. 7), as the IJV was collapsed all-through its neck course. Therefore, a correlation between CVPni and CVPi was only feasible in only 93 patients out of 106. By method 1, a highly significant positive correlation between CVPni and CVPi was found (ρ < 0.001) and a Pearson correlation coefficient (r = 0.913) for the studied patients (n = 93), while AUC for the ROC curve was 0.951 in the patients with CVPi ≥ 10cmH20. The mean CVPni for patients with CVPi < 10 cmH20 was 7.66 ± 1.20 cmH20 and it was 10.33 ± 1.09 cmH20 for patients with CVPi ≥ 10 cmH20 (Fig. 2a) ( Table 2).
For differentiating between patients with CVP < 10cmH20 and ≥ 10cmH20, method 1 had gained a sensitivity of 100%, a specificity of 79.31%, a positive predictive value of 74.47%, and a negative predictive value of 100% with an accuracy of 87.10% when compared to CVPi (Table 4).

For method 2 IJV area
The mean IJV cross-sectional area for patients with CVP < 10 cmH20 was 0.92 ± 0.31 cm 2 while it was 1.26 ± 0.37 cm 2 for patients with CVP ≥ 10 cmH20. There was a significant association between the IJV area and CVPi in both groups (p < 0.001).

CCA area
The mean CCA cross-sectional area for patients with CVP < 10 cmH20 was 0.54 ± 0.16 cm2, while for patients with CVP ≥ 10 cmH20 it was 0.51 ± 0.13 cm 2 . There was no significant association between the CCA cross-sectional area and CVPi (p = 0.222).
For the differentiation between patients with CVP < 10cmH20 and ≥ 10cmH20, method 2 had a sensitivity of 91.11%, a specificity of 85.48%, a positive predictive value of 82.00%, and a negative predictive value of 92.98% with an accuracy of 87.85% relative to CVPi ( Table 7).

Combination of both methods
For differentiating between patients with CVP < 10cmH20 and ≥ 10cmH20, the combination of both methods (methods 1 and 2) had gained a sensitivity of 88.57%, a specificity of 89.66%, a positive predictive value of 83.78%, and a negative predictive value of 92.86% with an accuracy of 89.25% as compared to CVPi (Table 8).

Discussion
The hemodynamic monitoring in the dialyzing patients during the dialysis sessions necessitates the CVP measurement. This is used to be done by central catheters in the IJV. The invasive nature of this procedure, especially in patients who do not have an indwelling catheter has directed the researchers for finding a non-invasive substitute.
The IJV US could be used for assessing the CVP and it has offered a simple and reliable alternative for the catheter measurement.
In this context, we used the IJV US for the CVP measurement in two different methods based on the literature review, including the collapsing point (method 1) of the IJV and the cross-sectional area measurements for the IJV and the CCA (method 2).
For method 1; our results are consistent with the Kerleroux et al. study, which had exclusively enrolled the hemodialysis patients like those in our study, but their sample size was much smaller than ours (22 patients) and they had reported a significant correlation between CVPni and CVPi with P < 0.0001 [10].
Siva et al. had also reported a highly significant positive correlation between CVPni and CVPi (ρ = 0.004) in their study population (44 patients) using method 1 [11].
Congruence with Xing et al. was also present, where they had used the same principle (in method 1) as ours for CVPni measurement. However, they used echocardiography in their patients for more accurate detection of the right atrium center instead of using the five cm additive estimation [12]. Despite being a more accurate method, but it would significantly decrease the merit of being a simpler and less time-consuming procedure, consequently, it requires more training for the operator when compared to the other methods used in our study and Kerleroux et al., Siva et al. studies [10,11]. Nonetheless, Xing et al. had also reported a similar significant positive correlation between CVPni and CVPi in both preoperative measurements (r = 0.90; ρ < 0.01) and in postoperative measurements (r = 0.93; ρ < 0.01) for their patients (118 patients) [12].
For method 2; our results are concordant with those of Hossein-Nejad et al. who performed their study on 52 non-ventilated patients and also reported a highly significant positive correlation between IJV/CCA crosssectional area ratio and CVPi (r = 0.728, p < 0.0001 at inspiration, and r = 0.736, p < 0.0001 at expiration), while the AUC for the ROC curve was 0.882 for predicting patients with CVPi < 10cmH20. They calculated a cut-off point (= 2) for the prediction of CVP ≥ 10cmH20, and they found a significant correlation between the IJV area and CVPi with no significant correlation between CCA area and CVPi [13]; however, our sample size is almost double theirs.
Bailey et al. had also documented similar results to ours. They had concluded that the IJV/CCA crosssectional ratio could predict the value of CVP. Their preliminary results suggested that if the IJV/CCA crosssectional area ratio was at least 2, then the CVP seemed to be ≥ 8 mmHg which is nearly close to 10 cmH20 (ρ < 0.001). It was a pilot study that was conducted in the pediatric burn population with a small sample size including only six patients [14].
Bano and Canuad had performed their study on 49 ventilated and non-ventilated patients. However, they measured IJV/CCA diameter ratio instead of cross-sectional area ratio and found a significant positive correlation between IJV/CCA diameter ratio and CVPi in only nonventilated patients at end-expiration (r = 0.439, n = 24, ρ = 0.032), and calculated a cuff-off (IJV/CCA) diameter ratio (= 1.75) for predicting CVP ≥ 10 cm H20 [15], which is matching with our results. However, they found no significant correlation between the IJV/CCA diameter ratio and CVP in non-ventilated patients at inspiration (r = 0.308, n = 24, p = 0.143) and in ventilated patients at both inspiration and expiration (r = 0.343, n = 25, p = 0.094 and r = 0.346, n = 25, p = 0.094, respectively). Our study was performed on non-ventilated patients and all our measurements were taken at end-expiration, so our results are concordant with theirs regarding their sub-group of non-ventilated patients who were being examined at end-expiration. Donahue et al. had utilized only the IJV diameter and IJV cross-sectional areas in both supine and 35° reverse Trendelenburg positions at both end-inspiration and end-expiration. They depicted a significant difference in IJV diameter in patients with a CVP < 10 or ≥ 10 cm H2O and a significant positive correlation was present between the IJV end-expiratory diameter and CVP (r = 0.82) in the supine position [16]. Their results are in line with ours as we also had found a significant correlation between the IJV area and CVPi (ρ < 0.001). A contradiction with the results of Elsadek et al. who had performed their study on smaller sample size (16 pediatrics patients) and had detected a poor correlation between IJV diameter or IJV area and the volume status (as predicted by left ventricular end-diastolic area) (LVEDA) [17]. However, the smaller sample size, the different age groups, and most importantly the use of LVEDA as a predictor of volume status instead of CVPi are all considered as probable causes for this contradiction.
Comparison and combination of both methods were then performed in terms of novelty; to our knowledge, our study was the first one to include and compare these two methods. Both methods show overall comparable accuracy for differentiation of patients with CVPi < 10cmH20 and ≥ 10cmH20 (87.10% for method 1 and 87.85% for method 2). Upon using both methods together for the same patient we had achieved a higher accuracy level approaching 89.25% which is better than using either method 1 or method 2 that had an accuracy of 87.10% and 87.85%, respectively. Thus, we propose the application of both methods together is recommended to improve the confidence in the acquired measurements; moreover, the experience with both methods could be beneficial when one of the two methods is not applicable or feasible for use in one patient.
It worth mentioning that one of the strengths of this study is the homogeneity of the studied sample (Nonventilated adult hemodialysis patients) but on the other hand, it is considered one of its limitations as it is not applicable for ventilated patients.
Some additional limitations were met in this work including.
For Method 1, its limitation was the inability to measure the extremes of CVP as no collapse point for the IJV was detected along its neck course. In our study, the exact measurements of CVPni could not be obtained in nine patients with very high CVPi (mean CVP for these 9 patients was 19.278 cmH20) as the IJV was distended all-through its neck course, moreover, the measurements were not feasible in another four patients who had a low CVPi (mean CVP for these 4 patients was 2.25cmH20) as the IJV was collapsed all-through its neck course. Nevertheless, this limitation is of little clinical importance because it could still detect if the CVP is low (below 5cmH20) or very high.
For Method 2, it only gives an estimate and not a direct reading of the CVP, however, in agreement with other studies [13,15,16]; we found that method 2 was able to accurately differentiate between CVPi < 10cmH20 and ≥ 10cmH20, and this is considered as an important issue during resuscitation of the critically ill patients. Further studies on larger samples are recommended to find out IJV/CCA cross-sectional area ratios that are well correlated with the exact CVPi values.
Finally, we recommend (from our experience) US measurement of CVP as the first method of choice in patients with a dialysis arteriovenous fistula, to avoid insertion of central venous catheters that could carry a risk of central venous thrombosis and deprives the patient of future central access by either Mahurkar or Permacath in case of a dysfunctional fistula.

Conclusion
The US of the IJV had offered a simple, feasible, cheap, reliable, and non-invasive technique that could be used for repeated monitoring of the CVP. The present study has a novelty of combining more than one US method and it had reported a higher accuracy of the acquired measurements as compared to the use of a single US method.