Survey
After exclusion of three lesions with undetectable vascularization patterns, online survey included 74 high-definition digital images of 37 lesions, and the questions below:
What is the pattern of vascularization in each picture?
Which light mode reveals vasculatures better?
The survey was distributed to a group of otolaryngologists who had none or limited experience in using image enhancement systems to define laryngeal vascular changes according to the ELS classification. We preferred a heterogeneous sampling method, also known as the maximum variation sampling method, as the purposive sampling technique as it captures a wide range of experience relating to the examination of the laryngeal lesions. The participants were selected from the whole population of otolaryngology residents and fellows in Turkey; the group comprised 12 residents in the minimum second year of their training program, 8 junior fellows within 10 years after termination of the training program, and 10 seniors with more than 10 years of work experience in general otolaryngology. The survey was sent to the participants via social media groups, with brief information about the ELS classification. In every case, the participants identified the predominant type of vascular pattern on the pictures and then selected the clearest picture in the pair.
Participants classified the vasculatures as pertaining to longitudinal or perpendicular patterns, being unaware of the authors’ evaluations and final pathological diagnoses. The evaluation of the images by the participants was compared to the evaluation of the authors to determine the congruence of the answers. The evaluation of the participants was correlated with their previous work experience in calendar years.
Statistics
Statistical analysis was performed using the Statistical Package for the Social Sciences version 25 software program (SPSS Inc., USA). A p value < 0.05 was considered statistically significant. The χ2 (chi square) statistic and a contingency table were used to evaluate the relevance between the vascularization patterns and the pathological diagnosis.
The congruence between the two groups of evaluators (the authors and the participants) was tested with using cross-tabulation, χ2 and Cohen’s Kappa estimates were determined. Intra-class correlation coefficient (ICC) analysis was performed to evaluate the inter-rater reliability between the participants. Each participant in the group evaluated every single photo separately; two-way random ICC was used. Bivariate Pearson correlation analysis was used to determine the correlation between participants’ evaluation and participants’ work experience.
RESULTS
The study included 40 patients with laryngeal lesions. The average age was 55.1 years (min 24 - max 76 years). In the group, 10% of the patients were female and 90% were male.
Full survey responses were received from 30 otolaryngologists, with a mean age of 35.73 years (min 26-max 60) and a minimum of 2 and maximum of 35 years’ work experience (mean: 10,56).