Search

To estimate the number of OCT scans necessary to detect moderate and rapid rates of retinal nerve fiber layer (RNFL) thickness worsening at different levels of accuracy using a large sample of glaucoma and glaucoma-suspect eyes.Descriptive and simulation study.Twelve thousand one hundred fifty eyes from 7392 adult patients with glaucoma or glaucoma-suspect status followed up at the Wilmer Eye Institute from 2013 through 2021. All eyes had at least 5 measurements of RNFL thickness on the Cirrus OCT (Carl Zeiss Meditec) with signal strength of 6 or more.Rates of RNFL worsening for average RNFL thickness and for the 4 quadrants were measured using linear regression. Simulations were used to estimate the accuracy of detecting worsening-defined as the percentage of patients in whom the true rate of RNFL worsening was at or less than different criterion rates of worsening when the OCT-measured rate was also at or less than these criterion rates-for two different measurement strategies: evenly spaced (equal time intervals between measurements) and clustered (approximately half the measurements at each end point of the period).The 75th percentile (moderate) and 90th percentile (rapid) rates of RNFL worsening for average RNFL thickness and the accuracy of diagnosing worsening at these moderate and rapid rates.The 75th and 90th percentile rates of worsening for average RNFL thickness were -1.09 μm/year and -2.35 μm/year, respectively. Simulations showed that, for the average measurement frequency in our sample of approximately 3 OCT scans over a 2-year period, moderate and rapid RNFL worsening were diagnosed accurately only 47% and 40% of the time, respectively. Estimates for the number of OCT scans needed to achieve a range of accuracy levels are provided. For example, 60% accuracy requires 7 measurements to detect both moderate and rapid worsening within a 2-year period if the more efficient clustered measurement strategy is used.To diagnose RNFL worsening more accurately, the number of OCT scans must be increased compared with current clinical practice. A clustered measurement strategy reduces the number of scans required compared with evenly spacing measurements.

To develop an internationally standardized and validated tool to assess skill in performing tube shunt surgery.A panel of 6 glaucoma surgeons developed a tool for assessing tube shunt surgery using a modified Dreyfus scale for skill acquisition. The tool was reviewed by a panel of 10 international content experts, and their comments were incorporated into the final rubric.A different panel of 8 international glaucoma specialists independently graded videos of surgical procedures performed by 6 surgeons at various levels of ophthalmic training.Inter-rater reliability for each step in the rubric was calculated.The tube shunt rubric contained 13 steps specific to tube shunt surgery and 7 global indices. The Cronbach α statistic, a measure of internal reliability, ranged from 0.75 to 0.97, indicating strong internal reliability for all 13 steps.The tube shunt assessment tool has face validity, content validity, and interobserver reliability, and can be used to assess tube shunt surgery skills. Further studies are required to determine predictive and construct validity.

To identify patterns of visual field (VF) loss based on unsupervised machine learning and to identify patterns that are associated with rapid progression.Cross-sectional and longitudinal study.A total of 2231 abnormal VFs from 205 eyes of 176 Ocular Hypertension Treatment Study (OHTS) participants followed over approximately 16 years.Visual fields were assessed by an unsupervised deep archetypal analysis algorithm and an OHTS-certified VF reader to identify prevalent patterns of VF loss. Machine-identified patterns of glaucoma damage were compared against those patterns previously identified (expert-identified) in the OHTS in 2003. Based on the longitudinal VFs of each eye, VF loss patterns that were strongly associated with rapid glaucoma progression were identified.Machine-expert correspondence and type of patterns of VF loss associated with rapid progression.The average VF mean deviation (MD) at conversion to glaucoma was -2.7 decibels (dB) (standard deviation [SD] = 2.4 dB), whereas the average MD of the eyes at the last visit was -5.2 dB (SD = 5.5 dB). Fifty out of 205 eyes had MD rate of -1 dB/year or worse and were considered rapid progressors. Eighteen machine-identified patterns of VF loss were compared with expert-identified patterns, in which 13 patterns of VF loss were similar. The most prevalent expert-identified patterns included partial arcuate, paracentral, and nasal step defects, and the most prevalent machine-identified patterns included temporal wedge, partial arcuate, nasal step, and paracentral VF defects. One of the machine-identified patterns of VF loss predicted future rapid VF progression after adjustment for age, sex, and initial MD.An automated machine learning system can identify patterns of VF loss and could provide objective and reproducible nomenclature for characterizing early signs of visual defects and rapid progression in patients with glaucoma.

Background It is essential to log resident-performed procedures to assess training programs and fulfill specialty requirements, but resident case numbers are often underreported. Current systems require inefficient data entry steps, and residents and fellows report that user interfaces and administrative burden contribute to logging inaccuracy. Objective To determine the accuracy, feasibility, and acceptability of a single logging approach for resident case logging. Methods In 2018, we implemented a case logging system integrated with the institutional electronic health record (EHR) and the Accreditation Council for Graduate Medical Education (ACGME) case log system to record procedures performed by ophthalmology residents. We compared the proportion of resident-performed cataract extractions in the EHR that were reported to ACGME for 3 periods: before the deployment of the new system (6 months), during the transition (6 months), and after the change (2 years). Resident satisfaction with the new system was evaluated using surveys. Results An analysis of resident cataract surgeries showed that the percentage of resident cases logged increased from 85% prior to implementation to 91% after implementation. The integrated system became the preferred case logging method, with 100% of all logged cases being entered using the new platform. Surveys showed that the percentage of trainees who were moderately or very satisfied with the case log process increased from 55% before implementation to 100% after implementation. Conclusions A resident case log system integrated with an EHR more accurately reflects resident operative volume and increases trainee satisfaction with the logging process.

The objective of this study was to develop an objective and easy-to-use glaucoma staging system based on visual fields (VFs).A total of 13,231 VFs from 8077 subjects were used to develop models and 8024 VFs from 4445 subjects were used to validate models.We developed an unsupervised machine learning model to identify clusters with similar VF values. We annotated the clusters based on their respective mean deviation (MD). We computed optimal MD thresholds that discriminate clusters with the highest accuracy based on Bayes minimum error principle. We evaluated the accuracy of the staging system and validated findings based on an independent validation dataset.The unsupervised k -means algorithm discovered 4 clusters with 6784, 4034, 1541, and 872 VFs and average MDs of 0.0 dB (±1.4: SD), -4.8 dB (±1.9), -12.2 dB (±2.9), and -23.0 dB (±3.8), respectively. The supervised Bayes minimum error classifier identified optimal MD thresholds of -2.2, -8.0, and -17.3 dB for discriminating normal eyes and eyes at the early, moderate, and advanced stages of glaucoma. The accuracy of the glaucoma staging system was 94%, based on identified MD thresholds with respect to the initial k -means clusters.We discovered that 4 severity levels based on MD thresholds of -2.2, -8.0, and -17.3 dB, provides the optimal number of severity stages based on unsupervised and supervised machine learning. This glaucoma staging system is unbiased, objective, easy-to-use, and consistent, which makes it highly suitable for use in glaucoma research and for day-to-day clinical practice.

PURPOSE: To assess risk factors contributing to corneal decompensation following glaucoma drainage device (GDD) implantation. DESIGN: Retrospective case control study. METHODS: Records of 1610 eyes that underwent GDD implantation between 06/01/2009 and 04/01/2020, at the Johns Hopkins Wilmer Eye Institute were reviewed. 79 eyes (5%) developed corneal decompensation, of which 46 underwent keratoplasty. These 79 cases were matched with 220 controls. Cox proportional hazard models with robust standard error estimates to account for clustering at the matched-pair level were used to assess risk factors for corneal decompensation. Kaplan-Meier Survival Analysis analyzed time to corneal decompensation. RESULTS: Mean (SD) age of cases and controls was 68 (12.3) and 60.5 (15.9) years, respectively. The mean time from GDD implantation to corneal decompensation was 32 months, and the cumulative probability of developing decompensation at 3, 6, and 9 years was 4.7%, 9.2%, and 14.8%, respectively. Final visual outcomes in cases were worse, with a final mean±SD visual acuity (LogMAR) of 1.96±1.25 relative to a mean±SD visual acuity of 1.11±1.36 in controls (p

In this prospective interventional case series that included 474 patients, there were no significant differences in visual field (VF) parameters between fields from patients tested one-at-a-time and simultaneously, except for fixation losses.To test for differences in reliability and performance parameters of patients taking VF tests while using a remote patient monitoring system to supervise 1 or 2 test sessions simultaneously.In a prospective interventional case series, 861 eyes of 474 consecutive patients undergoing automated perimetry during a 6-month period were monitored during the test using an audio/video-enabled remote monitoring system. Two patients were simultaneously tested (simultaneous test) by a single technician if they were ready for testing at the same time. Patients were otherwise tested individually (single test). Performance and reliability parameters including false negatives, false positives, fixation losses, mean deviation, pattern standard deviation, VF index, and test duration were compared between patients undergoing simultaneous tests and single tests. Patients undergoing remotely monitored testing, for whom a prior VF could be found, had performance and reliability parameters compared with those prior tests. VFs were analyzed separately for 2 test strategies: SITA Standard 24-2 and SITA Faster 24-2C.No significant parameter differences were observed among SITA Standard 24-2 VFs between single and simultaneous tests, except for fixation losses (single: 16.8±19.7%, simultaneous: 22.5±25.0%, P=0.01). Similarly, there were no significant differences observed among SITA Faster 24-2C tests. Paired analyses comparing remotely monitored VFs with prior traditionally monitored VFs showed no significant differences for any parameters, except for fewer fixation losses with remote monitoring (traditional: 23.6±27.5%, remote 17.7±20.8%, P=0.003).Remote patient monitoring of VF testing enabled technicians to supervise testing of 2 patients simultaneously with preserved performance and reliability.

Purpose To evaluate trends in glaucoma procedures in the United States Medicare population and to evaluate which physicians are performing newer procedures. Design Analysis of publicly available claims and payment data. Participants Surgeons and beneficiaries enrolled in United States Medicare between 1994 and 2017. Methods Data regarding payments to physicians by the Centers for Medicare and Medicaid Services (CMS) were downloaded for the years 2012 through 2017. Data regarding claims to CMS by physicians were requested and processed between 1994 and 2017. Procedure counts from both data sets then were normalized for changes in the Medicare population, with 1995 as the baseline. The normalized volumes of procedures over time were visualized, as were geographic distributions of surgeons and their volume of procedures. Main Outcome Measures Trends in procedure counts over time, geographic distribution of surgeons, and their volume of procedures. Results The number of trabeculectomies continues to decline and now is similar to the number of tubes. Use of the relatively new trabecular bypass shunts has increased rapidly. Surgeons performing these procedures are less likely to be performing traditional glaucoma surgeries as well. The number of laser-based cyclodestruction procedures increased after introduction of the endoscopic technique and again with the introduction of so-called micropulse procedures. The procedure counts obtained with physician payment data consistently are lower than those from claims data given the limitations of the payment data. Conclusions Glaucoma practice patterns change each time a new device or procedure is introduced. Collectively, the use of new microinvasive glaucoma surgery procedures has increased rapidly such that they now account for a significant majority of glaucoma surgeries. Given the almost complete lack of comparative data to inform surgeon choices regarding these procedures, it will be important that randomized studies are carried out to fill this gap.

Purpose The purpose of this study was to determine the rate of unplanned returns to the operating room (OR) within 180 days and at any time postoperatively after valved and non-valved tube shunt surgery. Design Retrospective case-control study. Methods A review of 357 eyes that underwent tube shunt surgery (151 valved, 206 non-valved) was conducted at an academic glaucoma service between January 2014 and December 2016. A control eye was time matched for each eye that underwent reoperation. Results The reoperation rate within 180 days was 16 of 151 (10.6%) for valved and 25 of 206 (12.1%) for non-valved tube shunts and at any time postoperatively was 31 of 151 (20.5%) for valved, and 47 of 206 (22.8%) for non-valved tube shunts. Mean postoperative follow-up was 2.8 ± 1.1 years. The most common reoperations within 180 days and at any time postoperatively after valved tube shunt surgery were tube revisions (43.8% within 180 days, 38.7% any time) and external cyclophotocoagulation (CPC) (31.3% within 180 days, 38.7% anytime). The most common reoperations within 180 days after non-valved tube shunt surgery were tube revisions (32.0%), external CPC (12.0%), and vitrectomy with anterior chamber washout (12.0%) and at any time postoperatively were tube revision (34.0%), external CPC (31.9%), and tube explant (12.8%). At last follow-up, eyes that returned to the OR and controls were similar in terms of mean intraocular pressure (IOP), proportion of eyes meeting target IOP, and change in visual acuity. Conclusions More than 20% of eyes undergoing tube shunt surgery returned to the OR at any time postoperatively with a mean follow-up of nearly 3 years, with more than 10% of eyes undergoing reoperation within the first 180 days. Rates of reoperation were similar between valved and non-valved tube shunts.

OBJECTIVE: To model the global test-retest variability of visual fields (VFs) in glaucoma. DESIGN: Retrospective cohort study. PARTICIPANTS: 8,088 VFs of 4,044 eyes from 4,044 participants. METHODS: We selected two reliable VFs (SITA 24–2) per eye measured with the Humphrey Field Analyzer within 30 days of each other. Each VF contained fixation losses (FL) ≤ 33%, false-negative rates (FNR) ≤ 20%, and false-positive rates (FPR) ≤ 20%. Stepwise linear regression was applied to select the model that best predicts the global test-retest variability from three categories of features of the first VF: (1) base parameters (age, mean deviation [MD], pattern standard deviation, glaucoma hemifield test, FPR, FNR, FL); (2) total deviation (TD) at each location; and (3) computationally-derived VF loss patterns (archetypes). The global test-retest variability was defined as root mean square deviation (RMSD) of TD values at all 52 VF locations. Model performance was assessed using adjusted R-squared and Bayesian information criterion (BIC). MAIN OUTCOME MEASURES: Archetype models to predict the global test-retest variability. RESULTS: The mean ± standard deviation of RMSD was 4.39 ± 2.55 dB. Between the two VF tests, TD values were more strongly correlated in central than in peripheral VF locations (intraclass coefficient range: 0.66–0.89; p < 0.001). Compared with the model using base parameters alone (adjusted R-squared = 0.45), adding TD values improved prediction accuracy of the global variability (adjusted R-(squared) = 0.53, p < 0.001) and BIC (decreased by 527; a change of > 6 represents strong improvement). Lower TD sensitivity in the outer-most peripheral VF locations was predictive of higher global variability. Adding archetypes to the base model improved model performance with an adjusted R-squared of 0.53 (p < 0.001) and lowering of BIC by 583. Greater variability was associated with concentric peripheral defect, temporal hemianopia, inferotemporal defect, near total loss, superior peripheral defect, and central scotoma (listed in order of decreasing statistical significance), and less normal VF and superior paracentral defect. CONCLUSIONS: Inclusion of archetype VF loss patterns and TD values based on first VFs improved the prediction of the global test-retest variability than using traditional global VF indices alone.

PURPOSE: Rule-based approaches to determining glaucoma progression from visual fields alone are discordant and have tradeoffs. To better detect when glaucoma progression is occurring, we utilized a longitudinal data set of merged VF and clinical data to assess the performance of a Convolutional Long Short-Term Memory (LSTM) neural network. DESIGN: Retrospective analysis of longitudinal clinical and visual field data. SUBJECTS: From two initial datasets of 672,123 visual fields from 213,254 eyes and 350,437 samples of clinical data, persons at the intersection of both datasets with four or more visual fields and corresponding baseline clinical data (cup-to-disc ratio, central corneal thickness, and intraocular pressure) were included. After exclusion criteria, specifically the removal of VFs with high false positive / negative rates and entries with missing data, were applied to ensure reliable data, 11,242 eyes remained. METHODS: Three commonly used glaucoma progression algorithms (Visual Field Index slope, Mean Deviation slope, and Pointwise Linear Regression) were used to define eyes as stable or progressing. Two machine learning models, one exclusively trained on visual field data and another trained on both visual field and clinical data, were tested. OUTCOME MEASURES: Area under the receiver operating characteristic (AUROC) curve and area under the precision-recall curve (AUPR) calculated on a held-out test set and mean accuracies from 3-fold cross validation were used to compare the performance of the machine learning models. RESULTS: The convolutional LSTM network demonstrated 91–93% accuracy with respect to the different conventional glaucoma progression algorithms given 4 consecutive visual fields for each subject. The model that was trained on both visual field and clinical data (AUROC between 0.89 and 0.93) had better diagnostic ability than a model exclusively trained on visual fields (AUROC between 0.79 and 0.82, p

Purpose: To describe the efficacy and safety of open versus closed conjunctival implantation of the XEN45 Gel Stent (Allergan Inc). Design: Retrospective, multicenter study. Participants: A total of 137 patients with glaucoma who underwent XEN45 implantation via open or closed conjunctival methods. The XEN45 was implanted as a stand-alone procedure or at the time of cataract surgery by 5 surgeons. Methods: Patient demographics, diagnoses, preoperative and postoperative clinical data, outcome measures including intraocular pressure (IOP), use of glaucoma medications, visual acuity, and complications were collected. Statistical analyses were performed with P < 0.05 as significant. Main Outcome Measures: Failure was defined as less than 20% reduction of IOP from medicated baseline or IOP >21 mmHg at 2 consecutive visits at postoperative month 1 and beyond, the need for subsequent operative intervention or additional glaucoma surgery, or a catastrophic event such as loss of light perception. Eyes that had not failed by these criteria and were not on glaucoma medications were considered complete successes. Eyes that had not failed but required glaucoma medications were defined as qualified successes. Results: Complete success was achieved in 31% and 56% of the closed and open groups, respectively (P = 0.01). Qualified success was achieved in 53% and 71% of the closed and open groups, respectively (P = 0.06). At postoperative month 12, the open conjunctiva group was using fewer glaucoma medications than the closed group (0.9 vs. 1.8, respectively; P = 0.02). At postoperative month 12, the open group had a significantly greater percentage of IOP reduction compared with the closed group (43.1% vs. 24.8%, respectively; P = 0.02). Postoperative needling rates were higher in the closed group compared with the open group (36.1% vs. 11.8%, P = 0.001). Conclusions: Implantation of the XEN45 with opening of the conjunctiva is a safe and efficacious procedure to lower IOP with comparable success rate and lower needling rate compared with the closed conjunctiva technique. Prospective evaluation of the various methods for XEN45 implantation will allow for further comparison.

In treated eyes with mild/suspect glaucoma, intraocular pressure (IOP) increments are associated with worsening mean deviation (MD) once IOP reaches the 20s. In moderate/advanced eyes, IOP increments are associated with worse visual field (VF) performance across the entire IOP range.The purpose of this study was to describe the relationship between mean treated IOP and VF worsening and understand how this relationship is affected by glaucoma severity and IOP range.A total of 1446 eyes of 869 treated glaucoma patients with at least 5 longitudinal reliable VF tests and IOP measures were included. Mixed-effects linear models were employed to understand the effect of eye-specific mean treated IOP on MD slope. Models included interaction terms to assess the differing relationships between MD slope and mean IOP by glaucoma severity (suspect/mild vs. moderate/advanced) and splines to account for the differing effects of mean IOP on MD slope at different IOP ranges (above or below 21 mm Hg).In suspect/mild glaucoma, when treated IOP values were21 mm Hg, a 1 mm Hg increment in IOP was not associated with an increase in the rate of VF worsening (P0.05) but when treated IOP values rose21 mm Hg, a 1 mm Hg increment in IOP was associated with faster VF worsening (-0.09 dB/y per 1 mm Hg increment, P0.05). In moderate/advanced disease, a 1 mm Hg increment in treated IOP was associated with faster VF worsening both below and above 21 mm Hg, but the effect was much more pronounced in the higher range (-0.02 dB/y per mm Hg increment21 mm Hg vs. -0.74 dB/y per mm Hg increment21 mm Hg, P0.05 for both).IOP remained associated with VF worsening in eyes with more advanced glaucoma throughout the IOP range but was only associated with VF worsening in eyes with less severe glaucoma at higher IOP.

Purpose To determine the unplanned return to operating room (OR) rate within 180 days and at any time postoperatively after trabeculectomy performed or supervised by an attending surgeon and to identify associated factors. Design Retrospective case-control study. Methods Review of 881 eyes that underwent trabeculectomy at an academic glaucoma service between January 2014 and December 2016. Inclusion criteria included adult patients with postoperative follow-up >180 days and no other glaucoma-related surgery within the prior year. For each eye that underwent reoperation, a control was time-matched within 1 month. Results The reoperation rate within 180 days was 9.5% (84/881) and at any time postoperatively was 23.3% (205/881). When intraoperative bleb needling cases were excluded, the reoperation rate was 6.5% (57/881) within 180 days and 13.6% (120/881) at any time postoperatively. Mean postoperative follow-up was 2.9 ± 1.1 years. The most common reoperations within 180 days were bleb revision (32.1%) and intraoperative bleb needling (28.6%) and at any time postoperatively were bleb needling (36.1%), bleb revision (23.9%), and tube shunt implant (13.2%). At last follow-up, eyes that returned to OR and control eyes were similar in terms of mean intraocular pressure (IOP), the proportion of eyes meeting target IOP, and change in visual acuity since the original trabeculectomy. Conclusions Nearly 10% of eyes returned to OR within 180 days and over 20% of eyes required reoperation at any time postoperatively with a mean follow-up of nearly 3 years. Additional studies should evaluate unplanned return to OR as a measure of surgical quality within ophthalmology.

Background/AimsTo identify objective criteria from optical coherence tomography (OCT) and perimetry that denote a useful, specific definition of glaucomatous optic neuropathy (GON) in eyes with open-angle glaucoma for comparisons among glaucoma research studies.MethodsA cross-sectional study of adult patients with glaucoma from nine centres on five continents evaluated de-identified physician diagnosis, OCT and perimetry results for 2580 eyes (1531 patients) in an online database. Each eye was graded by their glaucoma specialist as either definite, probable or not GON. Objective measures from OCT and perimetry, derived from an online consensus panel comprising 176 glaucoma specialists globally, were compared against the three diagnostic levels.ResultsDiagnoses were 54% ‘definite’, 22% ‘probable’ and 24% ‘not GON’. Using only OCT data or only field data had inadequate specificity (ConclusionsObjective criteria to define GON are practical and may be useful for comparisons among clinical studies to supplement subjective clinical assessment.

Purpose of review The purpose of this review is to describe the transition of ophthalmology education to a virtual curriculum during the COVID-19 pandemic. We highlight innovative solutions ophthalmic educators implemented for medical students, residents, and fellows. Recent findings Clinical and surgical ophthalmology volume abruptly decreased in March 2020 as a result of the COVID-19 public health crisis. All didactics, grand rounds, and journal clubs shifted to an online format. Ophthalmic educators collaborated with other institutions in the United States and abroad on shared virtual curricula and seminars, and simulation-based training tools were better emphasized for surgical skill acquisition. We share the latest literature written by ophthalmic educators on successful adaptation of these new instructional strategies. Summary The COVID-19 pandemic transformed ophthalmology education overnight and accelerated the adoption of innovative e-virtual learning resources for trainees. These changes will undoubtedly remain part of the new paradigm of medical education in the post-COVID era.

Two novel deep learning methods using a convolutional neural network (CNN) and a recurrent neural network (RNN) have recently been developed to forecast future visual fields (VFs). Although the original evaluations of these models focused on overall accuracy, it was not assessed whether they can accurately identify patients with progressive glaucomatous vision loss to aid clinicians in preventing further decline. We evaluated these 2 prediction models for potential biases in overestimating or underestimating VF changes over time.Retrospective observational cohort study.All available and reliable Swedish Interactive Thresholding Algorithm Standard 24-2 VFs from Massachusetts Eye and Ear Glaucoma Service collected between 1999 and 2020 were extracted. Because of the methods' respective needs, the CNN data set included 54 373 samples from 7472 patients, and the RNN data set included 24 430 samples from 1809 patients.The CNN and RNN methods were reimplemented. A fivefold cross-validation procedure was performed on each model, and pointwise mean absolute error (PMAE) was used to measure prediction accuracy. Test data were stratified into categories based on the severity of VF progression to investigate the models' performances on predicting worsening cases. The models were additionally compared with a no-change model that uses the baseline VF (for the CNN) and the last-observed VF (for the RNN) for its prediction.PMAE in predictions.The overall PMAE 95% confidence intervals were 2.21 to 2.24 decibels (dB) for the CNN and 2.56 to 2.61 dB for the RNN, which were close to the original studies' reported values. However, both models exhibited large errors in identifying patients with worsening VFs and often failed to outperform the no-change model. Pointwise mean absolute error values were higher in patients with greater changes in mean sensitivity (for the CNN) and mean total deviation (for the RNN) between baseline and follow-up VFs.Although our evaluation confirms the low overall PMAEs reported in the original studies, our findings also reveal that both models severely underpredict worsening of VF loss. Because the accurate detection and projection of glaucomatous VF decline is crucial in ophthalmic clinical practice, we recommend that this consideration is explicitly taken into account when developing and evaluating future deep learning models.

To examine the data quality and usability of visual acuity (VA) data extracted from an electronic health record (EHR) system during ophthalmology encounters and provide recommendations for consideration of relevant VA end points in retrospective analyses.Retrospective, EHR data analysis.All patients with eyecare office encounters at any 1 of the 9 locations of a large academic medical center between August 1, 2013, and December 31, 2015.Data from 13 of the 21 VA fields (accounting for 93% VA data) in EHR encounters were extracted, categorized, recoded, and assessed for conformance and plausibility using an internal data dictionary, a 38-item listing of VA line measurements and observations including 28 line measurements (e.g., 20/30, 20/400) and 10 observations (e.g., no light perception). Entries were classified into usable and unusable data. Usable data were further categorized based on conformance to the internal data dictionary: (1) exact match; (2) conditional conformance, letter count (e.g., 20/30Proportions of usable and unusable VA entries at the overall and subspecialty levels.All VA data from 513 036 encounters representing 166 212 patients were included. Of the 1 573 643 VA entries, 1 438 661 (91.4%) contained usable data. There were 1 196 720 (76.0%) exact match (category 1), 185 692 (11.8%) conditional conformance (category 2), 40 270 (2.6%) convertible conformance (category 3), and 15 979 (1.0%) plausible but not conformed entries (category 4). Visual acuity entries during visits with providers from retina (17.5%), glaucoma (14.0%), neuro-ophthalmology (8.9%), and low vision (8.8%) had the highest rates of unusable data. Documented VA entries with providers from comprehensive eyecare (86.7%), oculoplastics (81.5%), and pediatrics/strabismus (78.6%) yielded the highest proportions of exact match with the data dictionary.Electronic health record VA data quality and usability vary across documented VA measures, observations, and eyecare subspecialty. We proposed a checklist of considerations and recommendations for planning, extracting, analyzing, and reporting retrospective study outcomes using EHR VA data. These are important first steps to standardize analyses enabling comparative research.