Abstract
Purpose :
Infectious keratitis poses a severe threat to vision, particularly in regions where it reaches epidemic levels. Traditional diagnostic methods, such as corneal scraping cultures, often takes days or weeks, which delays crucial treatment to prevent vision loss. This delay is exacerbated in resource-limited areas with scarce culture availability, leading to empirical treatments with uncertain accuracy and significant misdiagnoses. This study aims to assess the accuracy of diagnosing infectious keratitis using photographs and patient history, aiming to bridge the diagnostic gap for more timely and effective interventions.
Methods :
We retrospectively analyzed 49 anterior biomicroscopy photographs from patients with infectious keratitis (65.3% female) caused by bacteria (57.1%), fungi (24.5%), and Acanthamoeba (18.4%) at the UNIFESP Cornea Division. Diagnostic benchmarks were established by cross-referencing images with data from the UNIFESP Microbiology Laboratory, including culture results. Images were presented in an online questionnaire, with each case evaluated twice: first with only the photograph and second with demographic and clinical details. Evaluators were categorized by expertise (ophthalmology residents, cornea fellows, or cornea specialists). Statistical analyses used Python 3.10.0, employing independent two-tailed T-tests and one-way ANOVA to compare groups (significance level set at 0.05).
Results :
No significant difference in diagnostic accuracy among evaluators was found: cornea specialists at 48% (SD 0.04), ophthalmology residents at 46% (SD 0.06), and cornea fellows at 48% (SD 0.01). In 64% of cases, evaluators changed their initial diagnosis upon reevaluation. Fatigue impact was examined across quartiles, revealing no significant differences (P=0.25). Diagnostic accuracy varied significantly based on information availability. When evaluating images alone, accuracy was 58% (SD 0.23), while considering both images and clinical details resulted in lower accuracy at 36% (P<0.001).
Conclusions :
Our study highlights diagnostic challenges in infectious keratitis, indicating a lower accuracy in comparison to previous research. This is the first study to include Acanthamoeba diagnosis into the assessment. These findings emphasize the urgent requirement for innovative, cost-effective diagnostic approaches to improve accuracy, particularly considering the restricted accessibility of corneal scraping cultures.
This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.