Impact of inter-vendor variability on evaluation of left ventricular mechanical dispersion
Vinesh Appadurai1,2, Gregory M Scalia1,2, Katherine Lau1,2
1Department of Cardiology, The Prince Charles Hospital, Queensland, Australia.
Purpose:
Left ventricular mechanical dispersion (LVMD) is a novel speckle tracking parameter for prognostic assessment of arrhythmic risk prediction. There is growing evidence to support its use in a variety of cardiomyopathic processes. There is paucity of data addressing any presence of inter-vendor discrepancies for LVMD. The aim of this study was to assess inter-vendor variability of LVMD in vendor specific software (VSS) and vendor independent software (VIS) in subjects with preserved and reduced left ventricular function.
Methods:
Fifty-nine subjects (14 normal subjects and 45 subjects with cardiac disease) were recruited and 2D speckle tracking echocardiographic images were acquired on two different ultrasound machines (GE and Philips). LVMD was measured by two different VSS (EchoPac GE and QLAB Philips) and one VIS (TomTec Arena).
Results:
There was significant bias and wide limits of agreement (LOA) in the overall cohort observed between two different VSS (17.6 ms; LOA: -29.6 to 64.8; r: .47). There was acceptable bias and narrower LOA with good agreement for LVMD between images obtained on different vendors when performed on VIS (-3.1 ms; LOA: -27.6 to 21.4; r: .75). QLAB LVMD was consistently higher than GE LVMD and TomTec LVMD in both preserved and reduced left ventricular function. LVMD measurements have high intra-vendor reproducibility with excellent inter and intra-observer agreement.
Conclusions:
There was acceptable bias and narrower LOA for LVMD assessment on a VIS. Inter-vendor variability exists for LVMD assessment between VSS. Serial measurements of LVMD should be performed using a single vendor for consistent and reliable results.
