Comparison between methods of vascular calibre characterization and measurement protocols: Influence of vessels number considered.

In recent years, retinal vascular calibre has emerged as an important non-invasive biomarker of microvascular health, widely used in ophthalmology for the detection and monitoring of various ocular diseases and has attracted growing interest in the study of various systemic diseases. However, significant methodological questions remain regarding the precise number of vessels to include in the calculation, as well as the agreement between the two most widely used formulas for estimating the retinal vascular calibre: the Parr-Hubbard formula and the Knudtson formula. The aim of this study is to provide evidence regarding these issues by analysing the agreement between both formulas and evaluating the influence of the number of vessels included in the calculation of retinal vascular calibre: central retinal artery equivalent (CRAE), central retinal vein equivalent (CRVE), and the arteriovenous ratio (AVR). Bland-Altman analysis, the interclass correlation coefficient (ICC), and the Pearson coefficient were used to analyse the accuracy and reliability of the measurements for different number of vessels. The results indicate that the agreement between both formulas improves as the number of analysed vessels decreases. Reducing the number of vessels analysed affects the accuracy of the CRAE and CRVE parameters calculated with both formulas. In contrast, the AVR calculated using the Parr-Hubbard formula remains more stable, even though the reproduction coefficient (RPC) and limits of agreement (LOA) values increase as the number of analysed vessels decreases. Therefore, using six or five vessels is recommended; if fewer are analysed, diagnostic decisions should be supported by additional tests and indicators.