Factors Associated With Laboratory Test Negativity Following a Transition in Specimen Collection in Microbial Keratitis Cases.

PURPOSE: Negative laboratory results make targeting microbial keratitis treatment difficult. We investigated factors associated with laboratory negativity in patients with microbial keratitis in the context of a transition to a new specimen collection method. METHODS: Microbial keratitis patients with associated laboratory tests were identified in the electronic health record of a tertiary care facility from August 2012 to April 2022. Patient demographics and laboratory results were obtained. Random sampling of 50% of charts was performed to assess the impact of the ocular history and pretreatment measures. The relationship between probability of negative laboratory results with demographics, ocular history, pretreatment measures, and utilization of a new specimen collection method (i.e. ESwab) was evaluated by multivariable logistic regression. RESULTS: Of 3395 microbial keratitis patients identified, 31% (n = 1051) had laboratory tests. Laboratory testing increased over time (slope = 2.5% per year, p < 0.001; 19.6% in 2013 to 42.2% in 2021). Laboratory negative rate increased over time (slope = 2.2% per year, p = 0.022; 48.5% in 2013 to 62.3% in 2021). Almost one-third of patients (31.2%, n = 164) were pretreated with steroids. Over two-thirds of patients were pretreated with antibiotics (69.5%, n = 367). 56.5% (n = 297) of patients were outside referrals. In multivariable regression, patients with corticosteroid pretreatment had lower odds of negative laboratory results (odds ratio [OR] = 0.49, p = 0.001). There were higher odds of negative laboratory results for every additional antibiotic prescribed to a patient prior to presentation (OR = 1.30, p = 0.006) and for specimens collected using ESwabs (OR = 1.69, p = 0.005). Age, prior eye trauma, outside referrals, and contact lens wear were not significantly associated with negative laboratory results. CONCLUSION: More microbial keratitis associated laboratory tests are being taken over time. Over 60% of tests were negative by 2022. Factors associated with negative laboratory test results included pretreatment with antibiotics and specimens collected with the new collection method.

A Comparison of Culture Results and Visual Acuity in Contact Lens Related Microbial Keratitis.

PURPOSE: Evaluate the effect of corneal and contact lens-related (CLR) culture results on visual acuity (VA) in patients with microbial keratitis (MK). METHODS: MK patients with corneal and CLR cultures were identified in the University of Michigan electronic health record from August 2012 to April 2022. Test results were classified as laboratory-positive or laboratory-negative. Linear regression was used to examine trends of VA and associations between changes in VA (differences of VA at 90-day and baseline VA) and corneal and CLR culture results, after adjustment for baseline VA. One-sample t-tests were used to test if the slope estimates were different from zero. RESULTS: MK patients (n = 50) were on average 49 years old (standard deviation = 20.9), 56% female, and 90% White. Positive corneal and CLR cultures were reported in 60% and 64% of patients, respectively, and 38% reported both. The agreement rate between corneal and CLR culture results was 30% (n = 15/50). LogMAR VA improved over time in patients with positive corneal and CLR cultures (Estimate=-0.19 per 10-day increase, p = 0.002), and in those with negative corneal and positive CLR cultures (Estimate= -0.17 per 10-day increase, p = 0.004). Compared to patients with negative corneal and CLR cultures, there was a trend toward improvement in VA for patients with positive corneal and CLR cultures (Estimate=-0.68, p = 0.068), and those with negative corneal and positive CLR cultures (Estimate= -0.74, p = 0.059), after adjusting for baseline VA. CONCLUSIONS: Positive CLR cultures are associated with significant improvement in VA over time. These additional cultures can provide guidance on appropriate antimicrobial selection, especially when corneal cultures are negative.

Automatic Classification of Slit-Lamp Photographs by Imaging Illumination.

PURPOSE: The aim of this study was to facilitate deep learning systems in image annotations for diagnosing keratitis type by developing an automated algorithm to classify slit-lamp photographs (SLPs) based on illumination technique. METHODS: SLPs were collected from patients with corneal ulcer at Kellogg Eye Center, Bascom Palmer Eye Institute, and Aravind Eye Care Systems. Illumination techniques were slit beam, diffuse white light, diffuse blue light with fluorescein, and sclerotic scatter (ScS). Images were manually labeled for illumination and randomly split into training, validation, and testing data sets (70%:15%:15%). Classification algorithms including MobileNetV2, ResNet50, LeNet, AlexNet, multilayer perceptron, and k-nearest neighborhood were trained to distinguish 4 type of illumination techniques. The algorithm performances on the test data set were evaluated with 95% confidence intervals (CIs) for accuracy, F1 score, and area under the receiver operator characteristics curve (AUC-ROC), overall and by class (one-vs-rest). RESULTS: A total of 12,132 images from 409 patients were analyzed, including 41.8% (n = 5069) slit-beam photographs, 21.2% (2571) diffuse white light, 19.5% (2364) diffuse blue light, and 17.5% (2128) ScS. MobileNetV2 achieved the highest overall F1 score of 97.95% (CI, 97.94%-97.97%), AUC-ROC of 99.83% (99.72%-99.9%), and accuracy of 98.98% (98.97%-98.98%). The F1 scores for slit beam, diffuse white light, diffuse blue light, and ScS were 97.82% (97.80%-97.84%), 96.62% (96.58%-96.66%), 99.88% (99.87%-99.89%), and 97.59% (97.55%-97.62%), respectively. Slit beam and ScS were the 2 most frequently misclassified illumination. CONCLUSIONS: MobileNetV2 accurately labeled illumination of SLPs using a large data set of corneal images. Effective, automatic classification of SLPs is key to integrating deep learning systems for clinical decision support into practice workflows.