intra- and interobserver agreement and relation to angiographic vasospasm and mortality
BACKGROUND: /st>Transcranial Doppler measurements of the middle cerebral artery flow velocity are widely used as an indicator of vasospasm after aneurysmal subarachnoid haemorrhage (SAH). We investigated inter- and intraoperator agreement in SAH patients and healthy volunteers using colour-coded transcranial Doppler (TCCD), with the secondary aim of describing prediction of angiographic vasospasm and mortality. METHODS: /st>Sixty patients and 70 healthy controls were each examined in duplicate by alternating operators. A total of 939 measurements divided on 201 examination sets were conducted by four observers. The Bland-Altman limits of agreement (LoA) were calculated using a variance components analysis. Angiography was performed on clinical indication and survival recorded at 30 days. RESULTS: /st>Differences between measurements increased with increasing average, and therefore, we analysed log-transformed values. Thus, LoA are given as ratios between measurements. There were no systematic intra- or interobserver differences (bias). The intraobserver LoA was 0.62-1.61 in patients and 0.67-1.50 in controls. However, they were 0.55-1.82 in patients with angiographic vasospasm, whereas in patients without, they were 0.66-1.52. The interobserver LoA was 0.55-1.81 in patients and 0.65-1.55 in controls, while in patients with and without angiographic vasospasm, they were 0.45-2.22 and 0.60-1.67, respectively. Flow velocity measurements day 6-10 were positively associated with 30 day mortality risk (P=0.02, logistic regression). CONCLUSIONS: /st>TCCD measurement variability is wider in patient measurements than in controls. This discrepancy can largely be explained by a higher degree of error in patients with angiographic vasospasm. Despite the considerable measurement variability in TCCD, values are predictive of outcome in SAH.
British Journal of Anaesthesia, 2013, Vol 110, Issue 4, p. 577-85