Reproducibility of qualitative assessment of stent struts coverage by optical coherence tomography

Assessment of stent strut coverage by optical coherence tomography (OCT) is not standardized. The methodology most commonly used is based on a visual binary qualitative assessment (strut covered or not). However, the influence of magnification (zoom setting) to the inter- and intra-observer agreements has not yet been evaluated. Aim of our study was therefore to evaluate the agreements of this approach, taking into account various zoom settings. 126 struts from 10 selected frames were independently evaluated by four observers using a stepwise approach increasing the zoom setting as following: (1) full view of the lumen (FV), (2) half view of the lumen (HV) and (3) a quarter view of the lumen (QV). Intra- and inter-observer agreements (κ) were assessed. The rate of uncoverage was determined for each strut as the number of times it was defined as uncovered divided by the total number of observations (maximum 12 = 3 zoom settings × 4 analysts) and expressed as percentage. The inter-observer κ values (mean [range]) were 0.32 [0.07-0.63], 0.40 [0.18-0.69] and 0.33 [0.09-0.6], within FV, HV and QV respectively. The intra-observer κ values were 0.60 [0.50-0.70], 0.75 [0.75-0.76] and 0.60 [0.50-0.70], within FV, HV and QV respectively. By increasing zoom setting the κ value of intra-observer agreement was 0.74 [0.58-0.83] (from FV to HV), 0.70 [0.56-0.83] (from HV to QV) and 0.70 [0.37-0.86] (from FV to QV). Overall, the rate of uncoverage was 15.5% [8.3-100%]. The OCT qualitative evaluation of strut coverage has wide inter and intra-observer agreements and is dependent of the zoom setting used during the analysis. A more reproducible approach would be needed to eventually increase the probability to link uncovered struts with clinical events.