Assessment of the Comparability of Pre- and Postoperative Eyelid Surgery Images Published on Social Media
Author: Santiago Ortiz-Perez
Base Hospital / Institution: Hospital Virgen de las Nieves / University of Granada
ePoster presentation
Abstract ID: 25-321
Purpose
To evaluate the technical comparability of pre- and postoperative eyelid surgery images published on Instagram.
Methods
500 image pairs were extracted from public accounts. Three independent observers (A, B, and C) evaluated each image based on four technical criteria: lighting, angle/position, filter/editing, and makeup/eyelash enhancements, using an ordinal scale (1 = different, 2 = similar, 3 = identical). Means, frequencies, Pearson and Spearman correlations, and Cohen’s Kappa coefficients were calculated.
Results
This preliminary analysis includes the first 200 image pairs. All 500 have been reviewed and full results will be available at presentation if accepted. Overall, images were poorly comparable. Observer A had a mean score of 1.60, B: 2.06, and C: 2.29. Lighting showed the least comparability (mean A: 1.28) but highest agreement between B and C (Kappa: 0.51). Spearman correlations were moderate (up to 0.61); Kappa values were low, especially for angle/position (A–C: 0.09). All three observers rated images as “identical” in 142 cases (17.75%). Pairwise agreement on score 3: A–B: 19.13%, A–C: 21.75%, B–C: 30.75%. Only 5.5% of images were rated fully identical across all parameters by at least two observers.
Conclusion
The analyzed images demonstrated significant limitations in terms of comparability, particularly with respect to lighting, angle, and editing. While there was some consistency in the overall scoring trends among the three observers, the low Kappa coefficients highlight the inherent subjectivity of this evaluation method. Standardizing both the capture and assessment of clinical images would be beneficial when publishing high-quality pre- and postoperative photographs. At a minimum, standardizing image capture is essential to provide patients with objective information and avoid shaping expectations based on non-comparable visuals. Artificial intelligence–based tools may offer greater objectivity and reproducibility, although their adoption entails technical and methodological challenges.
Additional Authors
| First name | Last name | Base Hospital / Institution |
|---|---|---|
| Ana María | Ropero-Molina | University of Granada |
| Alberto | Sanchez-Mellado | Hospital Virgen de las Nieves |
| Chakir | El-Mahraoui | Hospital Virgen de las Nieves |
| Francisco | Zamorano-Martin | Hospital Virgen de las Nieves |
