Transcranial Doppler ultrasound to assess cerebrovascular reactivity: reliability, reproducibility and effect of posture

Abstract

Transcranial Doppler ultrasound (TCD) allows measurement of blood flow velocities in the intracranial vessels, and can be used to assess cerebral vasodilator responses to a hypercapnic stimulus. The reliability of this technique has not been established, nor is there agreement about whether the technique should be performed in sitting or lying postures. We tested the intra- and inter-rater reliability of measures of cerebrovascular reactivity (CVR) in 10 healthy adults, in sitting and lying postures. Participants underwent triplicate bilateral ultrasound assessment of flow velocities in the middle cerebral arteries whilst sitting and lying supine prior to and during inhalation of Carbogen (5% CO2, 95% O2) for 2 min. This procedure was performed twice by each of two raters for a total of four sessions. CVR was calculated as the difference between baseline and the peak blood flow velocity attained during CO2 inhalation. Intraclass correlation coefficients (ICCs) for intra-rater reliability were greater sitting than lying for both raters (e.g. Rater 1 ICC sitting = 0.822, lying = 0.734), and inter-rater reliability was also greater in sitting (e.g. sitting ICC = 0.504, lying = 0.081). These results suggest that assessment of CVR using TCD should be performed with participants sitting in order to maximise CVR measurement reliability.

Keywords: Cerebrovascular reactivity; Doppler ultrasound; Posture; Reliability; Reproducibility.

Figure 1. Application of curve fitting to CBV data.

Beat-to-beat cerebral blood flow velocities (CBV) obtained during the 2-min period of hypercapnia were plotted and fitted with a spline curve, enabling the peak of the curve to be identified as peak CBV.

Figure 2. Intra-rater reliability correlations in sitting and lying, Rater 1.

Correlations between %CVR measures taken on two occasions by Rater 1 in sitting (left, ICC (3,1) = 0.822) and lying (right, ICC(3,1) = 0.734). Simple linear regression lines are shown, using the least squares method (Sigmaplot 12).

Figure 3. Intra-rater reliability correlations in sitting and lying, Rater 2.

Correlations between %CVR measures taken on two occasions by Rater 2 in sitting (left, ICC (3,1) = 0.463) and lying (right, ICC(3,1) = 0.015). Simple linear regression lines are shown, using the least squares method (Sigmaplot 12).

Figure 4. Inter-rater reliability comparisons showing superior reliability in sitting.

Correlations between %CVR measures obtained across both sessions by Rater 1 compared with Rater 2 for sitting (left, ICC(3,1) = 0.504) and lying (right, ICC(3,1) = 0.081). Simple linear regression lines are shown, using the least squares method (Sigmaplot 12).