Quantifying systemic congestion with Point-Of-Care ultrasound: development of the venous excess ultrasound grading system

This is a post-hoc analysis of data collected during a prospective cohort study at a tertiary cardiac surgery center from August 2016 to July 2017 (NCT02831907) [21]. Written consent was obtained for all patients and the project was approved by the Montreal Heart Institute Ethics Committee (2016-1946).

Non-critically ill patients 18 years and older undergoing cardiac surgery with the use of cardiopulmonary bypass were eligible to participate. Complete inclusion and exclusion criteria have been previously described [21]. Notably, patients with critical illness, AKI or delirium before surgery were excluded as well as patients with conditions that may have interfered with portal Doppler assessment (cirrhosis, portal thrombosis) and patients with severe chronic kidney disease (estimated glomerular filtration rate < 15 mL/min per 1.73 m² calculated using the Modified Diet in Renal Disease formula [25]) or dialysis.

All patients underwent repeated POCUS assessment the day before surgery, at ICU admission after surgery and daily from post-operative days 1 to 3. Each ultrasound assessment consisted of hepatic vein Doppler, portal vein Doppler, intra-renal venous Doppler and inferior vena cava (IVC) ultrasound. The complete method for hepatic, portal and renal Doppler assessment has been previously published [21]. The assessments were performed with concurrent electrocardiogram tracing to adequately identify the hepatic waveform phases during the cardiac cycle. Inter-observer variability for identifying portal and intra-renal vein Doppler patterns were good, as previously reported [21].

The IVC diameter was measured in its intra-hepatic portion at 2 cm of the junction with the hepatic veins using a longitudinal view from a sub-xiphoid position [26]. When the sub-xiphoid window was not appropriate the probe was moved laterally to the right side of the body, over the liver, until an adequate view was achieved. The maximal diameter during the respiratory cycle was measured.

Based on an original concept (P.R.), a multidisciplinary team composed of intensivists (A.D., P.R.), anesthesiologist (A.D.), emergency physicians (R.S., K.H.) and nephrologist (W.B.S) developed five VExUS grading system prototypes based on the severity of venous ultrasonographic markers (Fig. 1). For the hepatic vein Doppler, a systolic phase was of lesser amplitude than the diastolic phase but toward the liver was considered mild while the presence of a reversed systolic phase was considered severe [17, 27]. For the portal vein Doppler, a pulsatility fraction (PF) of 30–49% was considered mild while a PF > 50% was considered severe based on the previous studies [21, 28, 29]. For the intra-renal venous Doppler, a discontinuous pattern with a systolic and a diastolic phase was considered mild while a discontinuous pattern with only a diastolic phase was considered severe [18, 21]. The prototype grading systems were named VExUS “A” through “E” (Fig. 1) with multiple grades within each grading system. The VExUS score was determined for all patients and for all timepoints. Overall, echographic variables were > 95% complete within the dataset. In the case of a missing value for an echographic marker, the last known value for this marker for the patient (i.e., assessment performed the previous day) was imputed.

Demographic, baseline information, cumulative fluid balance information, hemodynamic parameters at the time of assessment and outcomes during hospitalization were all collected prospectively during the study as previously published [21]. The presence of left and right systolic ventricular dysfunction on trans-esophageal echocardiography performed by the attending anesthesiologist before cardiopulmonary bypass was also collected. Left ventricular dysfunction was defined as a left ventricular ejection fraction (LVEF) ≤ 40%, and right systolic ventricular dysfunction was defined by either tricuspid annular plane systolic excursion < 16 mm or right ventricular area of change < 35% [27]. Mean CVP measurements were obtained using a jugular central venous catheter. Measurements were noted before POCUS assessment at ICU admission and daily if the central line was still in place at the time of POCUS assessment. The mean CVP values were noted after verifying the position of the pressure transducer in relation with the patient. The height of the bed was adjusted if needed to insure proper positioning at the level of the right atrium (intersection between mid-axillary line and fourth intercostal space). N-terminal pro-beta natriuretic peptide (NT-pro-BNP) was also measured prospectively before surgery and on the morning of post-operative days 1, 2 and 3. The European System Operative Score Risk Evaluation score (EuroSCORE II) was calculated, as well as a validated risk score by Birnie et al. for the prediction of AKI in cardiac surgery patients based on pre-operative characteristics and validated for the Kidney Disease: Improving Global Outcomes (KDIGO) criteria [30]. During the post-operative period, AKI was defined by the KDIGO criteria as an increase of serum creatinine > 26 μmol/L within a 48‐h period or 50% from baseline creatinine within a week from cardiac surgery [31]. Vasopressors (norepinephrine, vasopressin, dopamine) and inotrope (epinephrine, milrinone, dobutamine) use was noted at the time of ultrasound assessment. The vasopressor–inotrope score (VIS) was calculated to estimate the degree of pharmacologic support at the time of ultrasound assessment [32, 33]. The VIS was calculated using the following formula: VIS = dopamine dose (μg/kg/min) + dobutamine dose (μg/kg/min) + (100 * epinephrine dose (μg/kg/min)) + (10 * milrinone dose (μg/kg/min)) + (10,000 * vasopressin dose (U/kg/min)) + (100 * norepinephrine dose (μg/kg/min)) [32, 34].

Results are presented in number (%) for dichotomous variables and in mean ± standard deviation (SD) or median and interquartile range (IQR) for continuous variables, where appropriate. Comparisons between two groups for continuous variables were done using Student t-test or Mann–Whitney U test, as appropriate, and comparison between two groups for categorical variables was done using Chi squared test. The prevalence of each VExUS grade was presented as descriptive data for each timepoint. The association between the VExUS grades and the risk of new‐onset of AKI was assessed using a Cox proportional hazards model with the VExUS grades considered as segmented time‐dependent covariates. After identifying the VExUS grade most associated with AKI in univariable analysis, multivariable Cox regression was performed. A first model was constructed by including the pre-operative risk of AKI as performed by Birnie et al. [30] as an a priori covariate. This score included age, sex, body mass index, smoker status, New York Heart Association functional status, diabetes, peripheral vascular disease, chronic hypertension, hemoglobin level, renal function, recent coronary angiogram, triple vessel disease, operative priority and procedure type. In addition, a second model was created by adding the VIS as a segmented time‐dependent covariate to the first model. As a sensitivity analysis, other multivariable models including CBP duration and cardiac output measured at the end of surgery were performed. Results are presented as hazard ratio (HR) with 95% confidence intervals (CI).

The sensitivity and specificity as well as the positive likelihood ratio (+LR) and the negative likelihood ratio (−LR) of the different VExUS grades assessed at ICU admission to predict AKI after cardiac surgery were presented. The same analysis was performed using different cut-offs of CVP (≥ 8, ≥ 10, ≥ 12 and ≥ 14 mmHg), as well as individual ultrasound findings included in the VExUS grading systems. The CVP cut-off used was chosen based on current literature and expert opinion [10, 11, 13, 35]. Results are presented with 95% CI. Leaf plots were created to visually compare the diagnostic performance of grade 3 of the VExUS C grading system and CVP ≥ 12 mmHg [36, 37]. Leaf plots were generated using an online tool [38]. As a supplementary analysis, specificity of grade 3 of the VExUS C grading system was compared to other variables using exact McNemar test for paired nominal data.

The association between the VExUS grading system prototypes and commonly used markers of venous congestion (cumulative fluid balance, NT-pro-BNP and CVP) and VIS was first assessed using generalized estimating equation models using a robust estimator for the covariance matrix and an exchangeable structure for the working correlation matrix was used. This type of analysis accounts for the repeated measures’ design, implying that the sample was not independent. In addition to the VExUS grades, the time of assessment (4 timepoints: Day 0 to Day 3) was included as a factor in the analysis and the interaction between the studied variable and the time of ultrasound assessment was tested. We found multiple significant interactions (p < 0.05) with the time of assessment. Consequently, we presented the association for each timepoint and for each VExUS grade. The difference between the different grades was assessed using one-way ANOVA or Kruskall–Wallis test, where appropriate depending on the distribution of data. In the presence of a significant result (p < 0.05), multiple post-hoc pairwise comparisons with Bonferroni correction were performed and significant results were presented. Statistical tests were performed in SPSS version 24 (IBM, Armonk, New York, USA).