[Modular semiquantitative quality assessment of ophthalmic health information on the internet--reproducibility and correlation between different assessment categories]. / Modulare semiquantitative Qualitätsbewertung augenheilkundlicher Gesundheitsinformationen im Internet--Reproduzierbarkeit und Korrelation zwischen verschiedenen Bewertungskategorien.

BACKGROUND: Quality evaluation of web-based health information in ophthalmology requires valid standards and reproducible assessment procedures. The objective was to evaluate the interrater-reliability of quality assessment of ophthalmic websites and a possible correlation between the results of the evaluation categories reliability/trustworthiness, quality of medical content and accessibility/usability. MATERIALS AND METHOD: After selection with the search engine "Google" 20 ophthalmic websites had been analysed by two independent evaluators using criteria checklists (modified Afgis transparency criteria, modified BITV test, medical content related to AMD) related to the aforementioned 3 categories. RESULTS: The interrater-reliability was almost perfectly estimated with Kappa-values of 0.91 for reliability/trustworthiness plus 0.89 for accessibility/usability and 0.79 for content. On average 62.5 % (+/- 17.43 %), 27.36 % (+/- 16.5 %) and 59.54 % (+/- 15.73 %) of the quality requirements were fulfilled for reliability/trustworthiness, content and accessibility/usability, respectively. No significant correlation was found between reliability and content (r = -0.039, p = 0.8709), reliability and accessibility/usability (r = -0.284; p = 0.228) plus content and accessibility/usability (r = 0.199; p = 0.4047). CONCLUSION: Sufficiently operationalised criteria are prerequisites for reproducible results of quality assessment of ophthalmic websites between different observers. The assessment within a single category, such as reliability/trustworthiness, does not allow one to draw conclusions on other categories such as content or accessibility/usability or the overall quality of a website. Therefore, simplified tools for quality assessment of health information by laymen and patients may have a limited validity.