Transcranial Doppler (TCD) ultrasound, a non-invasive technique for detecting cerebral blood flow velocity, has been widely employed clinically to evaluate cerebral vasospasm following subarachnoid haemorrhage (SAH) and to a lesser degree, to evaluate cerebral circulation under other medical conditions. However, being a highly observer dependent technique, intermittent TCD's diagnostic reliability has not been widely accepted. Current available data on the observer variability with TCD ultrasound are limited by the inappropriate use of the product moment correlation coefficient, the statistic used for assessing association rather than agreement. Continuous transcranial Doppler is thought to overcome the disadvantage of intermittent TCD technique. However, little information exists on time-related variability and reproducibility of continuous TCD signal.
Objectives: 1) Firstly, to quantify the intra- and inter-observer variability with the intermittent TCD in terms of the intraclass correlation coefficient (ICC) and define the confidence intervals. Secondly, to assess the impact of lack of practice with the technique on the observer variability and the accuracy of data generated. 2) To investigate the beat to beat variability and reproducibility of the continuous TCD signal over time (1 hour) in healthy volunteers and patients with aneurysmal subarachnoid haemorrhage.
Methods: The study was performed in 3 phases. In phase I, three observers insonated the middle cerebral artery (MCA) in 10 healthy volunteers and measured peak systolic and end diastolic cerebral blood flow velocities. The experiment was repeated on five healthy volunteers after an eight-week break during which none of the observers were allowed to use the TCD system. The ICC and the practice effect were quantified. In phase II, 10 healthy volunteers were studied using the continuous transcranial Doppler. The peak, diastolic, mean velocities and the pulsatility index from MCA were recorded continuously for 1 hour. For each subject the mean value, range and the coefficient of variation of each of the 3 velocity measurements were calculated. The beat to beat variability and reproducibility of the TCD signal was investigated. In phase III, 8 patients with aneurysmal SAH were examined in a similar fashion to phase II.
Results & Conclusion: In phase I, whilst a high level of observer agreement can be achieved with intermittent TCD monitoring (ICC up to 0.9), a practice effect was also documented. Lack of practice resulted in a reduction in the ICC to 0.67. Results of continuous (1 hour) TCD monitoring (phase II & III) revealed a marked fluctuation in the measurement of velocities in volunteers with broad ranges (up to 74-125 for peak velocity for one subject) and wide overall95% confidence intervals (up to 71-115 for peak velocity for grouped subjects). There was also a wide range of calculated beat to beat variability in each of the velocity measurements in both volunteers (from -24 to 58%) and patients (from -38 to 82%) and the extent of this variability is greater in patients as compared to controls (p<0.05). Intermittent measurements with the conventional TCD systems may potentially miss significant peaks or troughs in measurement. The integrity of the probe position over the duration of monitoring may also potentially influence the accuracy of data obtained with the continuous TCD systems. The study brings into question the previously established method of diagnosing the presence or absence of cerebral vasospasm based on intermittent TCD. Further studies particularly with larger patient populations are warranted.