The assessment of circulating volume using inferior vena cava collapse index and carotid Doppler velocity time integral in healthy volunteers: a pilot study

Thirty-four participants were recruited into the study. Data for one participant was not recorded correctly and was excluded from analysis leaving 33 (17 male, 16 female) data sets. The physiological changes observed in this study are summarised in Table 1.

PLR was associated with significant reduction in mean inferior vena cava collapsibility index from 0.24 to 0.17 (p < 0.01) – mean difference of -0.06 (95 % CI -0.10; -0.02). PLR was associated with a 23.6 % increase in mean stroke volume from 56.0 to 69.2 mL (p < 0.01), 18.2 % increase in cardiac output from 3.3 to 4.1 L/min (p < 0.001) – with a mean difference of 0.75 L/min (95 % CI 0.55; 0.96). Mean sniff collapsibility index reduced significantly from 0.57 to 0.42 (p < 0.01) – with a mean difference of -0.15 (95 % CI -0.25; -0.06). There was no significant change in carotid blood flow, from 770 mL/min to 773 mL/min (p = 0.8).

Results of the IVC collapsibility index are summarised in Fig. 2a. The IVC collapsibility index decreases on passive leg raise. One participant fully collapsed their IVC with a sniff with legs supine or raised.

Figure 2b shows PLR causes a significant rise in stroke volume from a median of 55 mL (IQR 49–63) to 69 mL (IQR 59–76). This demonstrates that stroke volume increases with passive leg raise as a consequence of increased central volume loading.

Figure 3 shows the correlation between IVCCI with legs supine, both with quiet respiration and sniff, and % change in stroke volume with PLR. The data shows that although IVCCI reduces and stroke volume increases with passive leg raise (corresponding to the central fluid loading), there is poor correlation between IVCCI and stroke volume changes – IVC collapsibility does not predict the subjects in which fluid redistribution improves stroke volume.

Passive leg raise showed no significant alteration in carotid artery blood flow rate in our volunteers (Fig. 4).

The change in stroke volume on PLR was plotted against the change in carotid VTI with passive leg raise (Fig. 5). This shows poor correlation between changes in stroke volume with changes in common carotid flow. Poor correlation was also observed when absolute values of stroke volume and carotid VTI with legs supine (r ² = 0.21) or raised (r ² = 0.29) were compared.

This study found that PLR was associated with a fall in IVCCI, increase in stroke volume and cardiac output, but no changes in carotid blood flow.

The magnitudes of IVCCI were greater with ‘sniff’ compared to quiet respiration, presumably due to greater changes in intrathoracic pressure. Mean reduction of IVCCI with PLR was similar with quiet respiration (29.2 %) and sniff (26.3 %) as a proportion of initial IVCCI.

We have shown that IVCCI, both with quiet respiration and ‘sniff’, decreases with PLR, suggesting that IVCCI alters with centralised blood volume and so may have potential to predict fluid responsiveness in spontaneously ventilating patients. However, we also found that the correlation between IVCCI and cardiac output was poor. The effect of PLR on IVCCI showed wide variation: ranging from a 96.6 % reduction in quiet respiration IVCCI with PLR and a 92.4 % reduction in sniff IVCCI with PLR to a decrease in IVCCI on PLR: opposite to the expected finding. Eight participants had a greater IVCCI post-PLR with quiet inspiration and five post-PLR with ‘sniff’ inspiration. These data suggest that variations in inspiratory force and volume may have a greater effect on IVCCI (and cardiac output) than volume centralisation with PLR in some subjects. The position change may affect respiratory effort and mechanics. Variability of IVCCI with breathing method has been reported by Kimura, who noted for the same inspiratory tidal volume diaphragmatic breathing caused greater IVCCI than chest wall breathing [21]. Thus IVCCI may have limited clinical potential unless inspiratory effort is standardised both between measurements and participants. Future work exploring the role of IVCCI should attempt to standardise respiratory effort.

Since our study was performed others have also explored the role of IVCCI and PLR. Corl [22] examined changes in IVCCI with PLR in spontaneously ventilating patients presenting to an adult emergency department, finding a mean IVCCI of 15.8 %, with IVCCI 4 min after PLR decreasing by an average of 0.5 %. The authors concluded IVCCI was a poor predictor of fluid responsiveness. This may also be consequent upon the 4-min delay between PLR and IVCCI measurement, as haemodynamic changes following PLR are transient as discussed further below. Panebianco [23] showed that IVC size and IVCCI are little changed when recorded supine or at 45 degrees, concluding that measurements taken supine and semi-upright may be regarded as equivalent. Lanspa examined IVCCI in 14 spontaneously ventilating ICU patients receiving 10 mL/kg intravenous fluid delivered over 20 min [24]. Fluid responsiveness was defined as increase in cardiac output of >15 %. Five subjects were fluid responders and nine were non-responders. An IVCCI of >15 % had a positive predictive value of 62 %, but excellent negative predictive value of 100 % (p = 0.03). Muller examined 40 spontaneously ventilating ICU patients with acute circulatory failure [25]. Muller found that IVCCI of >40 % had a positive predictive value of 72 % for fluid responsiveness (defined by increase in subaortic VTI of >15 % following infusion of 500 mL 6 % hydroxyethylstarch), but IVCCI of <40 % had a negative predictive value of 83 % and so could not exclude fluid responsiveness. Airapetian examined 59 spontaneously breathing ICU patients with acute circulatory failure [26]. This study found that an IVCCI of >42 % distinguished between fluid responders (>10 % increase in cardiac output with 500 mL 0.9 % saline infusion) and non-responders with high specificity (97 %) and positive predictive value (90 %) but low sensitivity.

In our study the changes in cardiac output measured by USCOM were short-lived and values returned to their pre-PLR values within around 30 s to 2 min. A recent study concurs; reporting that PLR results in a transient change in physiological measurements, with normal physiology restored within one to 4 min [27]. The time taken to restore to normal physiology was not an a priori measured variable in our study, as this rapid normalisation of physiology was not anticipated. However, this was noted consistently in all subjects, anecdotally, more rapidly in fitter and more athletic subjects.

The time to maximum change in stroke volume and return to baseline with PLR is not well defined. A study of healthy spontaneously ventilating volunteers by Delerme [6], measured changes in pulse oximetry waveforms with PLR, finding a significant difference between baseline and post-PLR, which were maintained for 5 min. Boulain reported haemodynamic changes persisting for 4 min post-PLR in mechanically ventilated ICU patients [28]. However, Lamia [27] notes that haemodynamic changes following PLR are transient so designed their study to record their data within 1 min of PLR. Best practice for performing a passive leg raise is discussed by Monnet [29], who notes that physiological effects from PLR may vanish after 1 min. This paper outlines how PLR should be optimally performed to assess for changes in stroke volume and discusses how differences in results in the literature may be due to differences in methods of PLR. The short duration of the PLR associated changes in stroke volume were not known at the time of conducting this study and consequently the exact times of each component of the study were not recorded. This may explain the range of changes in cardiac output reported here. The limited time window during which measurements should be taken potentially limits the utility of PLR without continuous cardiac monitoring.

Mean stroke volume increased 23.6 % from 56.0 to 69.2 mL. Mean cardiac output increased by 22.3 % from 3.3 L/min to 4.1 L/min with PLR, comparable to other studies [6, 26, 27]. A rise in cardiac output >10 % with 500 mL crystalloid bolus or PLR is commonly used to define fluid responsive patients [9, 10]. In our study 24 participants increased their cardiac output by >10 % on PLR and were therefore fluid responsive. Nine participants had changes in cardiac output of <10 %. This is surprising as all participants were expected fluid responsive as they had normal cardiovascular systems, were well and fasted for 8 h. Review of the quality of the suprasternal Doppler traces identified them as comparable to previous work and do not account for the non-responders. This finding may reflect the transient effect on cardiac output consequent upon PLR. The USCOM cardiac output readings in this study are lower than may be expected for healthy volunteers. A related study on this machine performed by a different operator within our institution also obtained lower than expected cardiac index readings [12]. A meta-analysis by Chong and Peyton [30] comparing USCOM with thermodilution noted that USCOM underestimates cardiac output by an average of 0.39 L/min with large variation between these methods of cardiac output measurement.

Our study reports that common carotid artery flow does not increase in response to PLR and we found no relationship between carotid artery blood flow and stroke volume, irrespective of leg positioning. This suggests that carotid VTI cannot be used as a surrogate measure for changes in stroke volume on PLR in healthy volunteers. This is contrary to a recent study by Marik where carotid artery blood flow differentiated fluid responders from non-responders [4]. In their intensive care unit based study of 34 patients, 17 non-responders to fluid boluses had minimal changes (mean increase 0.1 %) to carotid blood flow on PLR, while 17 fluid responders had a mean carotid blood flow increase of 79 %. This paper notes that work in their group on healthy volunteers found PLR caused an average 33 % increase in stroke volume and 16 % increase in carotid blood flow. However, there are differences in study design.

The Marik study measured the diameter of the carotid artery and found that this changed with fluid loading. Our study did not measure the carotid artery diameter, assuming no change with PLR, and thus our results may represent a false negative finding. Marik studied patients with septic shock in intensive care who had been partially or completely fluid resuscitated. Carotid blood flow changes in response to PLR maybe different in patients with sepsis as compared to healthy volunteers, possibly as a consequence of altered autoregulation [31]. As discussed, the changes in stroke volume consequent upon PLR are short lived and the carotid artery measurements were made at the end of our protocol, so delay in obtaining readings may result in recordings already returning to baseline. Gassner [20] also investigated the role of carotid artery ultrasound as a surrogate marker for cardiac output Their study on ICU patients found good correlation between carotid artery flow when compared with PA catheter (intra-class correlation 0.74) or arterial waveform pulse contour analysis (intra-class correlation 0.84) over a wide range of cardiac outputs.

This study reports an increase in stroke volume and fall in IVCCI with PLR in healthy volunteers. The wide range of IVCCIs reported with PLR suggests that the technique requires development prior to clinical use. In contrast to previous studies we report that common carotid artery blood flow does not change with PLR. The haemodynamic changes on the healthy volunteers following PLR were transient and rapidly returned to the baseline values, which may account for these findings. This may limit the role of PLR as an indicator of fluid responsiveness when using non-invasive monitors unable to provide continuous readings.