Abstract: Spaceflight associated neuro-ocular syndrome (SANS) refers to a distinct constellation of ocular, neurological and neuroimaging findings observed in astronauts during and following long duration spaceflight. These ocular findings, to include optic disc oedema, posterior globe flattening, chorioretinal folds and hyperopic shifts, were first described by NASA in 2011. SANS is a potential risk to astronaut health and will likely require mitigation prior to planetary travel with prolonged exposures to microgravity. While the exact pathogenesis of SANS is not completely understood, several hypotheses have been proposed to explain this neuro-ocular phenomenon. In this paper, we briefly discuss the current hypotheses and contributing factors underlying SANS pathophysiology as well as analogues used to study SANS on Earth. We also review emerging potential countermeasures for SANS including lower body negative pressure, nutritional supplementation and translaminar pressure gradient modulation. Ongoing investigation within these fields will likely be instrumental in preparing and protecting astronaut vision for future spaceflight missions including deep space exploration.

Abstract: Background/aims This retrospective multicentric panel study provides absolute numbers, types of and indications for corneal transplantation in Germany from 2011 to 2021 and sets them into the international context. Methods A questionnaire was sent to the 104 German ophthalmologic surgery departments and 93 (89%) provided their data. Results The number of reported keratoplasties more than doubled from 2011 (n=4474) to 2021 (n=8998). Lamellar keratoplasties (49% posterior (n=2883), 4% anterior (n=231)) surpassed penetrating keratoplasty (PKP, 47%, n=2721) for the first time in 2014. Since 2016, Descemet’s membrane endothelial keratoplasty (DMEK) has become the predominant keratoplasty procedure in Germany. Its number increased by 1.5-fold from 3850 (2016) to 5812 (2021). Main indications in 2021 were Fuchs’ endothelial corneal dystrophy (FECD, 43%), pseudophakic corneal decompensation (12%), repeated keratoplasty (11%), infections (7%), keratoconus (6%) and corneal scarring (4%, others: 9%). The PKP percentage decreased from 70.2% in 2011 (n=3141) to 31.7% in 2021 (n=2853). Descemet’s stripping (automated) endothelial keratoplasties (DSAEKs) decreased to 1% in 2021 (n=74). 98.6% of all posterior lamellar keratoplasties were DMEKs in Germany in 2021. The number of deep anterior lamellar keratoplasties (DALKs) remained comparable from 2011 (n=269) to 2021 (n=251, 2.8%). Conclusion Main indications for corneal transplantation in Germany (2021) were FECD and pseudophakic corneal decompensation. DMEK is by far the predominant corneal transplantation procedure since 2016 followed by PKP, whose absolute number decreased only slightly during the decade from 2011 to 2021. DALK proportions remain low, but stable, whereas DSAEK decreased annually and plays a minor role in Germany. Trial registration number NCT03381794 .

Abstract: Background Optical coherence tomography angiography (OCTA) enables fast and non-invasive high-resolution imaging of retinal microvasculature and is suggested as a potential tool in the early detection of retinal microvascular changes in Alzheimer’s Disease (AD). We developed a standardised OCTA analysis framework and compared their extracted parameters among controls and AD/mild cognitive impairment (MCI) in a cross-section study. Methods We defined and extracted geometrical parameters of retinal microvasculature at different retinal layers and in the foveal avascular zone (FAZ) from segmented OCTA images obtained using well-validated state-of-the-art deep learning models. We studied these parameters in 158 subjects (62 healthy control, 55 AD and 41 MCI) using logistic regression to determine their potential in predicting the status of our subjects. Results In the AD group, there was a significant decrease in vessel area and length densities in the inner vascular complexes (IVC) compared with controls. The number of vascular bifurcations in AD is also significantly lower than that of healthy people. The MCI group demonstrated a decrease in vascular area, length densities, vascular fractal dimension and the number of bifurcations in both the superficial vascular complexes (SVC) and the IVC compared with controls. A larger vascular tortuosity in the IVC, and a larger roundness of FAZ in the SVC, can also be observed in MCI compared with controls. Conclusion Our study demonstrates the applicability of OCTA for the diagnosis of AD and MCI, and provides a standard tool for future clinical service and research. Biomarkers from retinal OCTA images can provide useful information for clinical decision-making and diagnosis of AD and MCI.

Abstract: BACKGROUND Chat Generative Pre-trained Transformer (ChatGPT), a large language model by OpenAI, and Bard, Google's artificial intelligence (AI) chatbot, have been evaluated in various contexts. This study aims to assess these models' proficiency in the part 1 Fellowship of the Royal College of Ophthalmologists (FRCOphth) Multiple Choice Question (MCQ) examination, highlighting their potential in medical education. METHODS Both models were tested on a sample question bank for the part 1 FRCOphth MCQ exam. Their performances were compared with historical human performance on the exam, focusing on the ability to comprehend, retain and apply information related to ophthalmology. We also tested it on the book 'MCQs for FRCOpth part 1', and assessed its performance across subjects. RESULTS ChatGPT demonstrated a strong performance, surpassing historical human pass marks and examination performance, while Bard underperformed. The comparison indicates the potential of certain AI models to match, and even exceed, human standards in such tasks. CONCLUSION The results demonstrate the potential of AI models, such as ChatGPT, in processing and applying medical knowledge at a postgraduate level. However, performance varied among different models, highlighting the importance of appropriate AI selection. The study underlines the potential for AI applications in medical education and the necessity for further investigation into their strengths and limitations.

Abstract: Background Graves’ orbitopathy (GO) is subject to epidemiological and care-related changes. Aim of the survey was to identify trends in presentation of GO to the European Group On Graves’ Orbitopathy (EUGOGO) tertiary referral centres and initial management over time. Methods Prospective observational multicentre study. All new referrals with diagnosis of GO within September–December 2019 were included. Clinical and demographic characteristics, referral timelines and initial therapeutic decisions were recorded. Data were compared with a similar EUGOGO survey performed in 2012. Results Besides age (mean age: 50.5±13 years vs 47.7±14 years; p 0.007), demographic characteristics of 432 patients studied in 2019 were similar to those in 2012. In 2019, there was a decrease of severe cases (9.8% vs 14.9; p<0.001), but no significant change in proportion of active cases (41.3% vs 36.6%; p 0.217). After first diagnosis of GO, median referral time to an EUGOGO tertiary centre was shorter (2 (0–350) vs 6 (0–552) months; p<0.001) in 2019. At the time of first visit, more patients were already on antithyroid medications (80.2% vs 45.0%; p<0.001) or selenium (22.3% vs 3.0%; p<0.001). In 2019, the initial management plans for GO were similar to 2012, except for lid surgery (2.4% vs 13.9%; p<0.001) and prescription of selenium (28.5% vs 21.0%; p 0.027). Conclusion GO patients are referred to tertiary EUGOGO centres in a less severe stage of the disease than before. We speculate that this might be linked to a broader awareness of the disease and faster and adequate delivered treatment.

Abstract: Aims To assess the global burden and economic inequalities in the distribution of blindness and vision loss between 1990 and 2019. Methods A secondary analysis of the Global Burden of Diseases, Injuries and Risk Factors Study (GBD) 2019. Data for disability-adjusted life-years (DALYs) due to blindness and vision loss were extracted from the GBD 2019. Data for gross domestic product per capita were extracted from the World Bank database. Slope index of inequality (SII) and concentration index were computed to assess absolute and relative cross-national health inequality, respectively. Results Countries with high, high-middle, middle, low-middle and low Socio-demographic Index (SDI) had decline of age-standardised DALY rate of 4.3%, 5.2%, 16.0%, 21.4% and 11.30% from 1990 to 2019, respectively. The poorest 50% of world citizens bore 59.0% and 66.2% of the burden of blindness and vision loss in 1990 and 2019, respectively. The absolute cross-national inequality (SII) fell from −303.5 (95% CI −370.8 to −236.2) in 1990 to −256.0 (95% CI −288.1 to −223.8) in 2019. The relative inequality (concentration index) for global blindness and vision loss remained essentially constant between 1991 (−0.197, 95% CI −0.234 to −0.160) and 2019 (−0.193, 95% CI −0.216 to −0.169). Conclusion Though countries with middle and low-middle SDI were the most successful in decreasing burden of blindness and vision loss, a high level of cross-national health inequality persisted over the past three decades. More attention must be paid to the elimination of avoidable blindness and vision loss in low-income and middle-income countries.

Abstract: Background/aims To improve the accuracy of pterygium screening and detection through smartphones, we established a fusion training model by blending a large number of slit-lamp image data with a small proportion of smartphone data. Method Two datasets were used, a slit-lamp image dataset containing 20 987 images and a smartphone-based image dataset containing 1094 images. The RFRC (Faster RCNN based on ResNet101) model for the detection model. The SRU-Net (U-Net based on SE-ResNeXt50) for the segmentation models. The open-cv algorithm measured the width, length and area of pterygium in the cornea. Results The detection model (trained by slit-lamp images) obtained the mean accuracy of 95.24%. The fusion segmentation model (trained by smartphone and slit-lamp images) achieved a microaverage F1 score of 0.8981, sensitivity of 0.8709, specificity of 0.9668 and area under the curve (AUC) of 0.9295. Compared with the same group of patients’ smartphone and slit-lamp images, the fusion model performance in smartphone-based images (F1 score of 0.9313, sensitivity of 0.9360, specificity of 0.9613, AUC of 0.9426, accuracy of 92.38%) is close to the model (trained by slit-lamp images) in slit-lamp images (F1 score of 0.9448, sensitivity of 0.9165, specificity of 0.9689, AUC of 0.9569 and accuracy of 94.29%). Conclusion Our fusion model method got high pterygium detection and grading accuracy in insufficient smartphone data, and its performance is comparable to experienced ophthalmologists and works well in different smartphone brands.

Abstract: Aim To explore associations between artificial intelligence (AI)-based fluid compartment quantifications and 12 months visual outcomes in OCT images from a real-world, multicentre, national cohort of naïve neovascular age-related macular degeneration (nAMD) treated eyes. Methods Demographics, visual acuity (VA), drug and number of injections data were collected using a validated web-based tool. Fluid compartment quantifications including intraretinal fluid (IRF), subretinal fluid (SRF) and pigment epithelial detachment (PED) in the fovea (1 mm), parafovea (3 mm) and perifovea (6 mm) were measured in nanoliters (nL) using a validated AI-tool. Results 452 naïve nAMD eyes presented a mean VA gain of +5.5 letters with a median of 7 injections over 12 months. Baseline foveal IRF associated poorer baseline (44.7 vs 63.4 letters) and final VA (52.1 vs 69.1), SRF better final VA (67.1 vs 59.0) and greater VA gains (+7.1 vs +1.9), and PED poorer baseline (48.8 vs 57.3) and final VA (55.1 vs 64.1). Predicted VA gains were greater for foveal SRF (+6.2 vs +0.6), parafoveal SRF (+6.9 vs +1.3), perifoveal SRF (+6.2 vs −0.1) and parafoveal IRF (+7.4 vs +3.6, all p<0.05). Fluid dynamics analysis revealed the greatest relative volume reduction for foveal SRF (−16.4 nL, −86.8%), followed by IRF (−17.2 nL, −84.7%) and PED (−19.1 nL, −28.6%). Subgroup analysis showed greater reductions in eyes with higher number of injections. Conclusion This real-world study describes an AI-based analysis of fluid dynamics and defines baseline OCT-based patient profiles that associate 12-month visual outcomes in a large cohort of treated naïve nAMD eyes nationwide.

Abstract: Aims To evaluate the efficacy and safety of 0.01% atropine eye-drops in controlling myopia progression over 5 years. Methods Experimental, analytical, prospective, randomised and longitudinal study, in 361 right eyes from 361 children randomised into the control group (177 eyes without treatment) and treatment group (184 eyes with 0.01% atropine eye-drops). Children assigned to the treatment group used 0.01% atropine once a day every night and the control group’s children did not use any treatment or placebo. All the subjects completed an eye examination every 6 months for the 5 years of follow-up. The examination included subjective and objective refraction with cycloplegia, axial length (AL), keratometry and anterior chamber depth (ACD) to evaluate the efficacy of the treatment. It also included the anterior and posterior pole examination to evaluate the safety of the treatment. Results The SE increased −0.63±0.42D in children after 5 years of treatment with 0.01% atropine, while in the control group the increase was −0.92±0.56D. AL increased 0.26±0.28 mm in the treatment group compared with 0.49±0.34 mm in the control group. Atropine 0.01% showed an efficacy of 31.5% and 46.9% in the control of the SE and AL increase, respectively. ACD and keratometry did not have significant changes between groups. Conclusions Atropine 0.01% is effective in slowing myopia progression in a European population. There were no side effects after 5 years of 0.01% atropine.

Abstract: PROLOGUE: Glaucoma surgery has been, for many decades now, dominated by the universal gold standard which is trabeculectomy augmented with antimetabolites. Tubes also came into the scene to complement what we use to call conventional or traditional glaucoma surgery. More recently we experienced a changing glaucoma surgery environment with the "advent" of what we have become used to calling Minimally Invasive Glaucoma Surgery (MIGS). What is the unmet need, what is the gap that these newcomers aim to fill? Hippocrates taught us "bring benefit, not harm" and new glaucoma techniques and devices aim to provide safer surgery compared to conventional surgery. For the patient, but also for the clinician, safety is important. Is more safety achieved with new glaucoma surgery and, if so, is it associated with better, equivalent, or worse efficacy? Is new glaucoma surgery intended to replace conventional surgery or to complement it as an 'add-on' to what clinicians already have in their hands to manage glaucoma? Which surgery should be chosen for which patient? What are the options? Are they equivalent? These are too many questions for the clinician! What are the answers to the questions? What is the evidence to support answers? Do we need more evidence and how can we produce high-quality evidence? This EGS Guide explores the changing and challenging glaucoma surgery environment aiming to provide answers to these questions. The EGS uses four words to highlight a continuum: Innovation, Education, Communication, and Implementation. Translating innovation to successful implementation is crucially important and requires high-quality evidence to ensure steps forward to a positive impact on health care when it comes to implementation. The vision of EGS is to provide the best possible well-being and minimal glaucomainduced visual disability in individuals with glaucoma within an affordable healthcare system. In this regard, assessing the changes in glaucoma surgery is a pivotal contribution to better care. As mentioned, this Guide aims to provide answers to the crucial questions above. However, every clinician is aware that answers may differ for every person: an individualised approach is needed. Therefore, there will be no uniform answer for all situations and all patients. Clinicians would need, through the clinical method and possibly some algorithm, to reach answers and decisions at the individual level. In this regard, evidence is needed to support clinicians to make decisions. Of key importance in this Guide is to provide an overview of existing evidence on glaucoma surgery and specifically on recent innovations and novel devices, but also to set standards in surgical design and reporting for future studies on glaucoma surgical innovation. Designing studies in surgery is particularly challenging because of many subtle variations inherent to surgery and hence multiple factors involved in the outcome, but even more because one needs to define carefully outcomes relevant to the research question but also to the future translation into clinical practice. In addition this Guide aims to provide clinical recommendations on novel procedures already in use when insufficient evidence exists. EGS has a long tradition to provide guidance to the ophthalmic community in Europe and worldwide through the EGS Guidelines (now in their 5th Edition). The EGS leadership recognized that the changing environment in glaucoma surgery currently represents a major challenge for the clinician, needing specific guidance. Therefore, the decision was made to issue this Guide on Glaucoma Surgery in order to help clinicians to make appropriate decisions for their patients and also to provide the framework and guidance for researchers to improve the quality of evidence in future studies. Ultimately this Guide will support better Glaucoma Care in accordance with EGS's Vision and Mission. Fotis Topouzis EGS PresidentF1 CONTRIBUTORS: All contributors have provided the appropriate COI visible in detail at www.eugs.org/pages/guidesurgical/This manuscript reflects the work and thoughts of the list of individuals recognized above, but importantly, it reflects EGS views on the subject matter. Its strength originates from a team effort, where a cohesive group of authors and reviewers have worked towards a common goal and now stand behind the text in its entirety. The EGS nevertheless wishes to thank the following external contributors for their additional expertise, which was particularly valuable to the development of this Surgical Guide: Amanda Bicket, Jonathan Bonnar, Catey Bunce, Kuan Hu, Sheffinea Koshy, Jimmy Le, Tianjing Li, Francisco Otarola, Riaz Qureshi, Anupa Shah, Richard Stead and Marta Toth. A particular appreciation goes to Ian Saldanha for drafting the introductory overview on Core Outcomes on chapter 8. Finally, EGS would like to acknowledge Augusto Azuara Blanco, Chair of the Scientific and Guidelines Committee, for his expertise and advisory role throughout the entire process.Luis Abegao Pinto Editor Gordana Sunaric Mégevand Editor Ingeborg Stalmans Editor Luis Abegao Pinto, Centro Hospitalar Universitário Lisboa NorteHana Abouzeid, Clinical Eye Research Centre Adolph de Rothschild, AZ OphthalmologieEleftherios Anastasopoulos, Aristotle University of Thessaloniki, Papageorgiou Hospital, Thessaloniki, GreeceAugusto Azuara Blanco, Centre for Public Health, Queen's University BelfastLuca Bagnasco, Clinica Oculistica, DiNOGMI University of GenoaAlessandro Bagnis, Clinica Oculistica, IRCCS Ospedale Policlinico San MartinoJoao Barbosa Breda, Faculty of Medicine of the University of Porto, Porto, Portugal. Centro Hospitalar e Universitário São João, Porto, Portugal. KULeuven, BelgiumKeith Barton, University College London, Moorfields Eye HospitalAmanda Bicket, University of Michigan (Ann Arbor, MI, USA)Jonathan Bonnar, Belfast Health and Social Care TrustChiara Bonzano, Clinica Oculistica, IRCCS Ospedale Policlinico San MartinoRupert Bourne, Cambridge University HospitalAlain Bron, University Hospital DijonCatey Bunce, King's College LondonCarlo Cutolo, Clinica Oculistica, DiNOGMI University of Genoa, and IRCCS Ospedale Policlinico San MartinoBarbara Cvenkel, University Medical Centre Ljubljana Faculty of Medicine, University of LjubljanaAntonio Fea, University of TurinTheodoros Filippopoulos, Athens Vision Eye InstitutePanayiota Founti, Moorfields Eye Hospital NHS Foundation TrustStefano Gandolfi, U.O.C. Oculistica, University of Parma Julian Garcia Feijoo, Hospital Clinico San Carlos, Universidad Complutense, MadridGerhard Garhoefer, Medical University of Vienna, AustriaDavid Garway Heath, Moorfields Eye Hospital NHS Foundation Trust, London. Institute of Ophthalmology, University College London.Gus Gazzard, Moorfields Eye Hospital NHS Foundation Trust, London. Institute of Ophthalmology, University College London.Stylianos Georgoulas, Addenbrooke's, Cambridge University HospitalsDimitrios Giannoulis, Aristotle University of Thessaloniki, AHEPA Hospital, Thessaloniki, GreeceFranz Grehn, University Hospitals WuerzburgKuang Hu, NIHR Moorfields Biomedical Research Centre, London - Institute of Ophthalmology - University College LondonMichele Iester, Clinica Oculistica, DiNOGMI University of Genoa, and IRCCS Ospedale Policlinico San MartinoHari Jayaram, Moorfields Eye HospitalGauti Johannesson, Umea UniversityStylianos Kandarakis, National and Kapodistrian University of Athens, G. Gennimatas Hospital, Athens, Greece.Efthymios Karmiris, Hellenic Air Force General Hospital & National and Kapodistrian University of Athens, G. Gennimatas Hospital, AthensAlan Kastner, Clinica Oftalmologica Pasteur, Santiago, ChileAndreas Katsanos, University of Ioannina, GreeceChristina Keskini, Aristotle University of Thessaloniki, AHEPA HospitalAnthony Khawaja, Moorfields Eye Hospital and UCL Institute of OphthalmologyAnthony King, Nottingham University Hospitals NHS TrustJames Kirwan, Portsmouth hospitals university NHS trustMiriam Kolko, University of Copenhagen, Copenhagen University Hospital RigshospitaletSheffinea Koshy, University of GalwayAntoine Labbe, Quinze-Vingts -National Ophthalmology HospitalJimmy Le, Johns Hopkins Bloomberg School of Public Health, BaltimoreSanna Leinonen, Tays Eye Centre, Tampere University HospitalSophie Lemmens, University Hospitals UZ LeuvenTianjing Li, School of Medicine, University of Colorado Anschutz Medical Campus Giorgio Marchini, Clinica Oculistica, University Hospital, AOUI, Verona, ItalyJosé Martinez De La Casa, Hospital Clinico San Carlos. Universidad Complutense Andy McNaught, Gloucestershire Eye UnitFrances Meier Gibbons, Eye Center Rapperswil, SwitzerlandKarl Mercieca, University Hospitals Eye Clinic, Bonn, GermanyManuele Michelessi, IRCCS - Fondazione BiettiStefano Miglior, University of Milan BicoccaEleni Nikita, Moorfields Eye Hospital NHS Foundation TrustFrancesco Oddone, IRCCS -Fondazione BiettiFrancisco Otarola, Universidad de La FronteraMarta Pazos, Institute of Ophthalmology. Hospital Clínic Barcelona. Researcher at Institut d'Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS)Norbert Pfeiffer, Mainz University Medical CenterVerena Prokosh, University of Cologne, Center for ophthalmology.Riaz Qureshi, Johns Hopkins Medicine, BaltimoreGokulan Ratnarajan, Queen Victoria Hospital, East Grinstead, UKHerbert Reitsamer, University Clinic Salzburg / SALKLuca Rossetti, University of Milan, ASST Santi Paolo e Carlo, Milano, ItalyIan Saldanha, Johns Hopkins Bloomberg School of Public Health, BaltimoreCedric Schweitzer, CHU Bordeaux, Univ. Bordeaux, ISPED, INSERM, U1219 - Bordeaux Population Health Research Centre, France Andrew Scott, Moorfields Eye Hospital London Riccardo Scotto, Clinica Oculistica, DiNOGMI University of GenoaAnupa Shah, Queen's University BelfastGeorge Spaeth, Wills Eye Hospital/Sidney Kimmel Medical College/Thomas Jefferson UniversityIngeborg Stalmans, University Hospitals UZ Leuven, Catholic Univers

Abstract: Purpose To evaluate the potential of retinal optical coherence tomography (OCT) measurements and polygenic risk scores (PRS) to identify people at risk of cognitive impairment. Methods Using OCT images from 50 342 UK Biobank participants, we examined associations between retinal layer thickness and genetic risk for neurodegenerative disease and combined these metrics with PRS to predict baseline cognitive function and future cognitive deterioration. Multivariate Cox proportional hazard models were used to predict cognitive performance. P values for retinal thickness analyses are false-discovery-rate-adjusted. Results Higher Alzheimer’s disease PRS was associated with a thicker inner nuclear layer (INL), chorio-scleral interface (CSI) and inner plexiform layer (IPL) (all p<0.05). Higher Parkinson’s disease PRS was associated with thinner outer plexiform layer (p<0.001). Worse baseline cognitive performance was associated with thinner retinal nerve fibre layer (RNFL) (aOR=1.038, 95% CI (1.029 to 1.047), p<0.001) and photoreceptor (PR) segment (aOR=1.035, 95% CI (1.019 to 1.051), p<0.001), ganglion cell complex (aOR=1.007, 95% CI (1.002 to 1.013), p=0.004) and thicker ganglion cell layer (aOR=0.981, 95% CI (0.967 to 0.995), p=0.009), IPL (aOR=0.976, 95% CI (0.961 to 0.992), p=0.003), INL (aOR=0.923, 95% CI (0.905 to 0.941), p<0.001) and CSI (aOR=0.998, 95% CI (0.997 to 0.999), p<0.001). Worse future cognitive performance was associated with thicker IPL (aOR=0.945, 95% CI (0.915 to 0.999), p=0.045) and CSI (aOR=0.996, 95% CI (0.993 to 0.999) 95% CI, p=0.014). Prediction of cognitive decline was significantly improved with the addition of PRS and retinal measurements. Conclusions and relevance Retinal OCT measurements are significantly associated with genetic risk of neurodegenerative disease and may serve as biomarkers predictive of future cognitive impairment.

Abstract: Aim To investigate if active learning of contrast sensitivity (CS) in bilateral age-related macular degeneration (AMD) correlates better than visual acuity (VA) with vision-related quality of life (VRQoL) using factor analysis-calibrated National Eye Institute Visual Function Questionnaire-25 (NEI VFQ-25). Methods Prospective cross-sectional observational study in 93 patients (186 eyes) with bilateral AMD. CS was measured in one eye at a time with the quantitative CS function (qCSF) method (Adaptive Sensory Technology). Same-day VRQoL was assessed with factor analysis-calibrated NEI VFQ-25 visual function and socioemotional scales. Mixed-effects multiple linear regression analyses evaluated the associations of the qCSF outcomes and VA with the NEI VFQ-25 scales. A subgroup analysis on patients with AMD with VA more than 20/25 in both eyes was performed. Results Compared with VA, CS outcomes were associated with larger effect on both visual function scale (standardised beta coefficients (β*) for area under the logarithm of CSF (AULCSF) curve and CS thresholds at 1.5, 3 and 6 cycles per degree (cpd): β*=0.50, 0.48, 0.52, 0.46, all p<0.001, respectively, vs β*=−0.45 for VA, all p<0.001) and socioemotional scale (β* for AULCSF and CS threshold at 6 cpd: β*=0.44, 0.44 vs β*=−0.42 for VA, all p<0.001). In patients with AMD with VA more than 20/25 in both eyes (N=20), both VFQ-25 scales and all CS outcomes were significantly reduced. Conclusions qCSF-measured CS strongly correlates with patient-reported VRQoL in bilateral AMD, even stronger than VA does. This study further validates qCSF-measured CS as a promising functional endpoint for future clinical trials in AMD.

Abstract: Background/aims This study evaluates the performance of the Airdoc retinal artificial intelligence system (ARAS) for detecting multiple fundus diseases in real-world scenarios in primary healthcare settings and investigates the fundus disease spectrum based on ARAS. Methods This real-world, multicentre, cross-sectional study was conducted in Shanghai and Xinjiang, China. Six primary healthcare settings were included in this study. Colour fundus photographs were taken and graded by ARAS and retinal specialists. The performance of ARAS is described by its accuracy, sensitivity, specificity and positive and negative predictive values. The spectrum of fundus diseases in primary healthcare settings has also been investigated. Results A total of 4795 participants were included. The median age was 57.0 (IQR 39.0–66.0) years, and 3175 (66.2%) participants were female. The accuracy, specificity and negative predictive value of ARAS for detecting normal fundus and 14 retinal abnormalities were high, whereas the sensitivity and positive predictive value varied in detecting different abnormalities. The proportion of retinal drusen, pathological myopia and glaucomatous optic neuropathy was significantly higher in Shanghai than in Xinjiang. Moreover, the percentages of referable diabetic retinopathy, retinal vein occlusion and macular oedema in middle-aged and elderly people in Xinjiang were significantly higher than in Shanghai. Conclusion This study demonstrated the dependability of ARAS for detecting multiple retinal diseases in primary healthcare settings. Implementing the AI-assisted fundus disease screening system in primary healthcare settings might be beneficial in reducing regional disparities in medical resources. However, the ARAS algorithm must be improved to achieve better performance. Trial registration number NCT04592068.

Abstract: Objective To evaluate the role of automated optical coherence tomography (OCT) segmentation, using a validated deep-learning model, for assessing the effect of C3 inhibition on the area of geographic atrophy (GA); the constituent features of GA on OCT (photoreceptor degeneration (PRD), retinal pigment epithelium (RPE) loss and hypertransmission); and the area of unaffected healthy macula. To identify OCT predictive biomarkers for GA growth. Methods Post hoc analysis of the FILLY trial using a deep-learning model for spectral domain OCT (SD-OCT) autosegmentation. 246 patients were randomised 1:1:1 into pegcetacoplan monthly (PM), pegcetacoplan every other month (PEOM) and sham treatment (pooled) for 12 months of treatment and 6 months of therapy-free monitoring. Only participants with Heidelberg SD-OCT were included (n=197, single eye per participant). The primary efficacy endpoint was the square root transformed change in area of GA as complete RPE and outer retinal atrophy (cRORA) in each treatment arm at 12 months, with secondary endpoints including RPE loss, hypertransmission, PRD and intact macular area. Results Eyes treated PM showed significantly slower mean change of cRORA progression at 12 and 18 months (0.151 and 0.277 mm, p=0.0039; 0.251 and 0.396 mm, p=0.039, respectively) and RPE loss (0.147 and 0.287 mm, p=0.0008; 0.242 and 0.410 mm, p=0.00809). PEOM showed significantly slower mean change of RPE loss compared with sham at 12 months (p=0.0313). Intact macular areas were preserved in PM compared with sham at 12 and 18 months (p=0.0095 and p=0.044). PRD in isolation and intact macula areas was predictive of reduced cRORA growth at 12 months (coefficient 0.0195, p=0.01 and 0.00752, p=0.02, respectively) Conclusion The OCT evidence suggests that pegcetacoplan slows progression of cRORA overall and RPE loss specifically while protecting the remaining photoreceptors and slowing the progression of healthy retina to iRORA.

Abstract: Background/aims To report a first-in-human trial in open-angle glaucoma (OAG) subjects treated with a new microinterventional biostent-reinforced cyclodialysis technique to enhance supraciliary aqueous drainage. Methods Subjects (N=10; 74.1±7.9 years old) with OAG and cataracts underwent combined phacoemulsification cataract surgery with implantation of a permanent endoscleral supraciliary biostent to reinforce a controlled cyclodialysis cleft. The biostent comprised decellularised scleral allograft tissue microtrephined into a polymer tubular implant intraoperative/postoperative safety, intraocular pressure (IOP) and glaucoma medications were tracked through 12 months postimplantation. Results Baseline medicated IOP averaged 24.2±6.9 mm Hg with subjects using 1.3±0.8 IOP-lowering medications. Successful biostent implantation was achieved in all individuals without significant complications. Immediate IOP lowering was sustained through 1 year. Twelve-month mean IOP was reduced 40% from baseline to 14.6±3.2 mm Hg (p=0.004; paired two-tailed t-test), and 80% of patients achieved >20% IOP reduction. Biostenting reduced glaucoma medication use 62%, from a baseline mean of 1.3 required medications to 0.5 medications (p=0.037) at postoperative 12 months. The biotissue implant was well tolerated and demonstrated good endothelial safety with only 11% endothelial cell loss at 12 months after combined phaco-biostenting surgery, similar to that expected after phacoemulsification alone. Mean BCVA increased from baseline 20/130 Snellen to 20/36 at postoperative 12 months (p=0.001). Conclusion Supraciliary biostenting in OAG patients is well tolerated, has a good safety profile and produces long-term IOP-lowering while reducing glaucoma medication requirements.

Abstract: Aims To identify potential risk factors for myopia in children and adolescents and assess the credibility of each evidence, providing reference for the development of myopia prevention strategies. Methods We searched PubMed, Web of Science and Embase databases from inception to April 2022 to find systematic reviews or meta-analyses investigating the relationship between potential risk factors and myopia, and conducted an umbrella review. We recalculated the pooled effect sizes and 95% CIs of each potential risk factor through random-effects model, and reported its 95% prediction interval and between-study heterogeneity. Small-study effect and excess of significance bias were assessed to reveal potential publication bias. Results Twelve publications were included in this umbrella review, including eight meta-analyses and four qualitative systematic reviews. Twenty-two factors were identified, of which 16 were analysed quantitatively. Ten factors showed statistically significant association with myopia. Myopia in one or two parents and per additional hour of time spend outdoors per week were rated as highly suggestive evidence. Near work and gender were evaluated as suggestive evidence. The other five factors are weak evidence. Conclusions We found several risk factors for myopia with different levels of evidence, of which parental myopia presented the strongest association with myopia in children and adolescents. Our findings contribute to a better understanding of the association between potential risk factors and myopia among children and adolescents and are important for informing parenting, education, clinical practice guidelines and public health policy. PROSPERO TRIAL REGISTRATION NUMBER CRD42022333053.

Abstract: Aims To assess the efficacy and safety of a standardised hyperbaric oxygen therapy protocol (HBOT) monitored by fluorescein angiography (FA) in patients with retinal artery occlusion (RAO). Methods It is a prospective, non-comparative, monocentric study conducted between July 2016 and March 2022. All consecutive patients diagnosed with RAO within 7 days underwent visual acuity measurement, FA, macular optical coherence tomography (OCT) and OCT-angiography. They received two daily HBOT sessions (2.5 atmosphere absolute, 90 min) until revascularisation assessed by FA. Complete ophthalmic follow-up was scheduled at day 14, day 21 and at 1 month. The main outcome measure was a best-corrected visual acuity (BCVA) improvement defined as a decrease ≥0.3 logMAR at 1 month. Results Thirty-one patients were included and received a mean number of 33.9 (13–56) HBOT sessions. Retinal revascularisation was observed in 48.4% and 87.1% of patients at days 14 and 21, respectively. The mean BCVA on referral and at 1 month was 1.51 logMAR and 1.10 logMAR, respectively. Fifteen (48.4%) patients achieved the main outcome measure. Six (19.4%) patients experienced minor barotrauma that did not require HBOT discontinuation. The univariate analysis showed that antiplatelet-treated patients (p=0.044) and patients with a poor initial BCVA (p=0.008) were more likely to achieve a BCVA improvement. OCT-angiography was not sensitive enough to diagnose RAO or assess revascularisation. Conclusion In RAO patients monitored by FA until spontaneous revascularisation of the central retinal artery, HBOT was effective and safe.

Abstract: Aim To predict antivascular endothelial growth factor (VEGF) treatment requirements, visual acuity and morphological outcomes in neovascular age-related macular degeneration (nAMD) using fluid quantification by artificial intelligence (AI) in a real-world cohort. Methods Spectral-domain optical coherence tomography data of 158 treatment-naïve patients with nAMD from the Fight Retinal Blindness! registry in Zurich were processed at baseline, and after initial treatment using intravitreal anti-VEGF to predict subsequent 1-year and 4-year outcomes. Intraretinal and subretinal fluid and pigment epithelial detachment volumes were segmented using a deep learning algorithm (Vienna Fluid Monitor, RetInSight, Vienna, Austria). A predictive machine learning model for future treatment requirements and morphological outcomes was built using the computed set of quantitative features. Results Two hundred and two eyes from 158 patients were evaluated. 107 eyes had a lower median (≤7) and 95 eyes had an upper median (≥8) number of injections in the first year, with a mean accuracy of prediction of 0.77 (95% CI 0.71 to 0.83) area under the curve (AUC). Best-corrected visual acuity at baseline was the most relevant predictive factor determining final visual outcomes after 1 year. Over 4 years, half of the eyes had progressed to macular atrophy (MA) with the model being able to distinguish MA from non-MA eyes with a mean AUC of 0.70 (95% CI 0.61 to 0.79). Prediction for subretinal fibrosis reached an AUC of 0.74 (95% CI 0.63 to 0.81). Conclusions The regulatory approved AI-based fluid monitoring allows clinicians to use automated algorithms in prospectively guided patient treatment in AMD. Furthermore, retinal fluid localisation and quantification can predict long-term morphological outcomes.

Abstract: Aims To compare the long-term outcomes of deep anterior lamellar keratoplasty (DALK) with penetrating keratoplasty (PK) in keratoconus. Methods Retrospective comparative case series (228 DALKs and 274 PKs). A biphasic linear model was used to describe the postoperative outcome of the endothelial cell density (ECD). Visual acuity, specular microscopy, corneal topography and optical coherence tomography findings were recorded. Results Graft survival of the 502 keratoconus eyes was 96.7 at 10 years and 95.6% at 20 years. Visual acuity improved from 20/378±5.1 lines preoperatively to 20/32±2.1 lines at 30 months. The corneal ECD decreased from 2494±382 cells/mm2 to 1521±659 cells/mm2 at 10 years. The mean simulated keratometry increased from 44.88±2.54 D at 1 year to 46.60±3.0 D at 3 years. The mean follow-up was 103.4 months for DALKs and 106.1 months for PKs. The cumulated incidence of postoperative ocular hypertension requiring treatment was significantly higher in PKs than in DALKs. The early- and late-phase rates of ECD loss were significantly lower in DALKs than in PKs. These figures in DALKs were 50% of those observed in PKs. The simulated mean keratometry was significantly higher in DALKs than in PKs in the mid but not in the long term. No significant differences in visual acuity were observed between both groups. Manual dissection-DALK featured slower visual recovery than PK and big bubble-DALK, whereas big bubble-DALK and PK featured similar visual recovery. Conclusions DALK featuring higher endothelial survival and lower risk of postoperative ocular hypertension may be superior to PK when indicated for keratoconus.

Abstract: Background/aims Defocus incorporated multiple segments (DIMS) spectacle lenses are known to be able to inhibit axial length (AL) growth in myopic children compared with single vision (SV) spectacle lenses. However, it is not known whether AL growth is sufficiently inhibited to achieve the treatment goal of physiological AL growth. Methods Of the data already collected in 2014–2017 by Lam et al , the AL growth with DIMS and SV spectacle lenses was re-evaluated according to the age-matched myopia control system. The individual AL growth after the first year of treatment of each eye was plotted against the corresponding age of the same time point in a colour-coded scheme. The two treatment groups were further subdivided based on their age and their baseline AL. Results Overall, 65% (61% of male, 70% of female) of eyes with DIMS spectacle lenses and 16% (16% of male, 16% of female) of eyes with SV spectacle lenses are within range of physiological AL growth rate. Median AL growth rate of eyes with DIMS spectacle lenses is also within the range of physiological growth. In the subgroups, eyes with DIMS spectacle lenses were also superior to the ones with SV spectacle lenses regarding this treatment goal. Of the children with SV spectacle lenses, older children and children with eyes with high baseline AL were least likely to achieve physiological AL growth rate. Conclusions DIMS spectacle lenses can bring the AL growth rate of myopic children to the level of physiological AL growth rate, indicating 100% reduction of excessive myopic AL growth, independent of age and baseline AL. Older children and children with eyes with high AL have the risk to have increased AL growth without treatment.

Abstract: BACKGROUND/AIMS Socioeconomic status (SES) is associated with late disease presentation and poorer outcomes. We evaluate the effect of SES on treatment outcomes and report the correlation between SES and baseline characteristics of participants in the Treatment of Advanced Glaucoma Study. METHODS Pragmatic multicentre randomised controlled trial. Four hundred and fifty-three patients presenting with advanced open-angle glaucoma in at least one eye (Hodapp-Parrish-Anderson classification). Participants were randomised to either glaucoma drops (medical arm) or trabeculectomy (surgery arm). Clinical characteristics, Quality of life measurement (QoL) and SES defined by the Index of Multiple Deprivation are reported. Subgroup analysis explored treatment effect modifications of SES at 24 months. Correlation between SES and baseline characteristics was tested with the χ2 test of association for dichotomous variables and pairwise Pearson's correlation for continuous variables. RESULTS The mean visual field mean deviation was -17.2 (6.7)dB for the most deprived quintile of participants and -13.0 (5.5) for the least deprived quintile in the index eye. At diagnosis, there was a strong correlation between SES and ethnicity, age, extent of visual field loss and number of visits to opticians prior to diagnosis. At 24 months, there was no evidence that the treatment effect was moderated by SES. CONCLUSIONS In patients presenting with advanced glaucoma. SES at baseline is correlated with poorer visual function, poorer Visual Function Questionnaire-25 QoL, ethnicity, age and number visits to an optician in the years preceding diagnosis. SES at baseline does not have an effect of the success of treatment at 24 months. TRIAL REGISTRATION NUMBER ISRCTN56878850.

Abstract: Background/aims Deep learning systems (DLSs) for diabetic retinopathy (DR) detection show promising results but can underperform in racial and ethnic minority groups, therefore external validation within these populations is critical for health equity. This study evaluates the performance of a DLS for DR detection among Indigenous Australians, an understudied ethnic group who suffer disproportionately from DR-related blindness. Methods We performed a retrospective external validation study comparing the performance of a DLS against a retinal specialist for the detection of more-than-mild DR (mtmDR), vision-threatening DR (vtDR) and all-cause referable DR. The validation set consisted of 1682 consecutive, single-field, macula-centred retinal photographs from 864 patients with diabetes (mean age 54.9 years, 52.4% women) at an Indigenous primary care service in Perth, Australia. Three-person adjudication by a panel of specialists served as the reference standard. Results For mtmDR detection, sensitivity of the DLS was superior to the retina specialist (98.0% (95% CI, 96.5 to 99.4) vs 87.1% (95% CI, 83.6 to 90.6), McNemar’s test p<0.001) with a small reduction in specificity (95.1% (95% CI, 93.6 to 96.4) vs 97.0% (95% CI, 95.9 to 98.0), p=0.006). For vtDR, the DLS’s sensitivity was again superior to the human grader (96.2% (95% CI, 93.4 to 98.6) vs 84.4% (95% CI, 79.7 to 89.2), p<0.001) with a slight drop in specificity (95.8% (95% CI, 94.6 to 96.9) vs 97.8% (95% CI, 96.9 to 98.6), p=0.002). For all-cause referable DR, there was a substantial increase in sensitivity (93.7% (95% CI, 91.8 to 95.5) vs 74.4% (95% CI, 71.1 to 77.5), p<0.001) and a smaller reduction in specificity (91.7% (95% CI, 90.0 to 93.3) vs 96.3% (95% CI, 95.2 to 97.4), p<0.001). Conclusion The DLS showed improved sensitivity and similar specificity compared with a retina specialist for DR detection. This demonstrates its potential to support DR screening among Indigenous Australians, an underserved population with a high burden of diabetic eye disease.

Abstract: Background Data on population-based self-reported dual vision and hearing impairment are sparse in Europe. We aimed to investigate self-reported dual sensory impairment (DSI) in European population. Methods A standardised questionnaire was used to collect medical and socio-economic data among individuals aged 15 years or more in 29 European countries. Individuals living in collective households or in institutions were excluded from the survey. Results Among 296 677 individuals, the survey included 153 866 respondents aged 50 years old or more. The crude prevalence of DSI was of 7.54% (7.36–7.72). Among individuals aged 60 or more, 9.23% of men and 10.94% of women had DSI. Eastern and southern countries had a higher prevalence of DSI. Multivariable analyses showed that social isolation and poor self-rated health status were associated with DSI with ORs of 2.01 (1.77–2.29) and 2.33 (2.15–2.52), while higher income was associated with lower risk of DSI (OR of 0.83 (0.78–0.89). Considering country-level socioeconomic factors, Human Development Index explained almost 38% of the variance of age-adjusted prevalence of DSI. Conclusion There are important differences in terms of prevalence of DSI in Europe, depending on socioeconomic and medical factors. Prevention of DSI does represent an important challenge for maintaining quality of life in elderly population.

Abstract: Background/aims The purpose of this study was to investigate retinal structure in detail of subjects with autosomal-dominant (AD) and autosomal-recessive (AR) PROM1-associated retinal degeneration (PROM1-RD), study design: institutional, cross-sectional study. Methods Four eyes from four subjects (three with AD and one with AR) PROM1-RD were investigated by ophthalmic examination including best-corrected visual acuity (BCVA) and multimodal retinal imaging: fundus autofluorescence (FAF), spectral-domain optical coherence tomography (SD-OCT) and adaptive optics scanning light ophthalmoscopy. Quantitative assessment of atrophic lesions determined by FAF, thickness of individual retinal layers and cone photoreceptor quantification was performed. Results BCVA ranged from 20/16 to 20/200. Initial pathological changes included the presence of hyperautofluorescent spots on FAF imaging, while later stages demonstrated discrete areas of atrophy. In all patients, thinning of the outer retinal layers on SD-OCT with varying degrees of atrophy could be detected depending on disease-causing variants and age. Cone density was quantified both in central and/or at different eccentricities from the fovea. Longitudinal assessments were possible in two patients. Conclusions PROM1-RD comprises a wide range of clinical phenotypes. Depending on the stage of disease, the cone mosaic in PROM1-RD is relatively preserved and can potentially be targeted by cone-directed interventions.

Abstract: Aims To investigate myopic maculopathy in Chinese children with high myopia and its association with choroidal and retinal changes. Methods This cross-sectional study included Chinese children aged 4–18 years with high myopia. Myopic maculopathy was classified by fundus photography and retinal thickness (RT) and choroidal thickness (ChT) in the posterior pole were measured by swept-source optical coherence tomography. A receiver operation curve was used to determine the efficacy of fundus factors in classifying myopic maculopathy. Results In total, 579 children aged 12.8±3.2 years with a mean spherical equivalent of −8.44±2.20 D were included. The proportions of tessellated fundus and diffuse chorioretinal atrophy were 43.52% (N=252) and 8.64% (N=50), respectively. Tessellated fundus was associated with a thinner macular ChT (OR=0.968, 95% CI: 0.961 to 0.975, p<0.001) and RT (OR=0.977, 95% CI: 0.959 to 0.996, p=0.016), longer axial length (OR=1.545, 95% CI: 1.198 to 1.991, p=0.001) and older age (OR=1.134, 95% CI: 1.047 to 1.228, p=0.002) and less associated with male children (OR=0.564, 95% CI: 0.348 to 0.914, p=0.020). Only a thinner macular ChT (OR=0.942, 95% CI: 0.926 to 0.959, p<0.001) was independently associated with diffuse chorioretinal atrophy. When using nasal macular ChT for classifying myopic maculopathy, the optimal cut-off value was 129.00 µm (area under the curve (AUC)=0.801) and 83.85 µm (AUC=0.910) for tessellated fundus and diffuse chorioretinal atrophy, respectively. Conclusion A large proportion of highly myopic Chinese children suffer from myopic maculopathy. Nasal macular ChT may serve as a useful index for classifying and assessing paediatric myopic maculopathy. Trial registration number NCT03666052.

Abstract: Background Deep learning (DL) is promising to detect glaucoma. However, patients’ privacy and data security are major concerns when pooling all data for model development. We developed a privacy-preserving DL model using the federated learning (FL) paradigm to detect glaucoma from optical coherence tomography (OCT) images. Methods This is a multicentre study. The FL paradigm consisted of a ‘central server’ and seven eye centres in Hong Kong, the USA and Singapore. Each centre first trained a model locally with its own OCT optic disc volumetric dataset and then uploaded its model parameters to the central server. The central server used FedProx algorithm to aggregate all centres’ model parameters. Subsequently, the aggregated parameters are redistributed to each centre for its local model optimisation. We experimented with three three-dimensional (3D) networks to evaluate the stabilities of the FL paradigm. Lastly, we tested the FL model on two prospectively collected unseen datasets. Results We used 9326 volumetric OCT scans from 2785 subjects. The FL model performed consistently well with different networks in 7 centres (accuracies 78.3%–98.5%, 75.9%–97.0%, and 78.3%–97.5%, respectively) and stably in the 2 unseen datasets (accuracies 84.8%-87.7%, 81.3%-84.8%, and 86.0%–87.8%, respectively). The FL model achieved non-inferior performance in classifying glaucoma compared with the traditional model and significantly outperformed the individual models. Conclusion The 3D FL model could leverage all the datasets and achieve generalisable performance, without data exchange across centres. This study demonstrated an OCT-based FL paradigm for glaucoma identification with ensured patient privacy and data security, charting another course toward the real-world transition of artificial intelligence in ophthalmology.

Abstract: Background Evidence on the practical application of artificial intelligence (AI)-based diabetic retinopathy (DR) screening is needed. Methods Consented participants were screened for DR using retinal imaging with AI interpretation from March 2021 to June 2021 at four diabetes clinics in Rwanda. Additionally, images were graded by a UK National Health System-certified retinal image grader. DR grades based on the International Classification of Diabetic Retinopathy with a grade of 2.0 or higher were considered referable. The AI system was designed to detect optic nerve and macular anomalies outside of DR. A vertical cup to disc ratio of 0.7 and higher and/or macular anomalies recognised at a cut-off of 60% and higher were also considered referable by AI. Results Among 827 participants (59.6% women (n=493)) screened by AI, 33.2% (n=275) were referred for follow-up. Satisfaction with AI screening was high (99.5%, n=823), and 63.7% of participants (n=527) preferred AI over human grading. Compared with human grading, the sensitivity of the AI for referable DR was 92% (95% CI 0.863%, 0.968%), with a specificity of 85% (95% CI 0.751%, 0.882%). Of the participants referred by AI: 88 (32.0%) were for DR only, 109 (39.6%) for DR and an anomaly, 65 (23.6%) for an anomaly only and 13 (4.73%) for other reasons. Adherence to referrals was highest for those referred for DR at 53.4%. Conclusion DR screening using AI led to accurate referrals from diabetes clinics in Rwanda and high rates of participant satisfaction, suggesting AI screening for DR is practical and acceptable.

Abstract: Aims To determine endophthalmitis rates and risk factors following intraocular surgeries in the Medicare population. Methods Retrospective, cross-sectional study from 2016 to 2019 in Medicare fee-for-service and Medicare Advantage beneficiaries. 100% Medicare claims were used to identify eyes that underwent intraocular surgery. Exclusion criteria included eyes that had intraocular surgery 42 days before or after and eyes in patients with a history of endophthalmitis within 12 months. Results 9 744 400 intraocular surgeries were performed among Medicare beneficiaries from 2016 to 2019. A 42-day postoperative endophthalmitis rate was 0.09%. Endophthalmitis rate by surgery type was 0.43% for corneal transplant, 0.36% for secondary intraocular lens (IOL), 0.24% for retina, 0.16% for glaucoma, 0.11% for cataract combined with other procedures and 0.08% for cataract surgeries alone. On multivariable analysis, the risk of endophthalmitis was increased for all intraocular surgery types when compared with cataract surgeries; adjusted OR (aOR) 5.30 (p<0.01) for corneal transplant, aOR 4.50 (p<0.01) for secondary IOL, aOR 3.00 (p<0.01) for retina, aOR 1.93 (p<0.01) for glaucoma, aOR 1.45 (p<0.01) for combined cataract surgeries. Increased risk of developing endophthalmitis was associated with older age (≥85 vs 65–75 years: aOR 1.36; p<0.01), male sex (aOR 1.20; p<0.001) and greater Charlson Comorbidity Index (≥7 vs 0: aOR 1.79; p<0.01). Conclusions Postoperative endophthalmitis rate after intraocular surgeries was 0.09% between 2016 and 2019 for Medicare beneficiaries. Endophthalmitis rates were highest for corneal transplant surgeries followed by secondary IOL surgeries and lowest for cataract surgeries. Older age, male gender and higher comorbidity index were associated with risk of endophthalmitis.

Abstract: Aims To develop an algorithm to classify multiple retinal pathologies accurately and reliably from fundus photographs and to validate its performance against human experts. Methods We trained a deep convolutional ensemble (DCE), an ensemble of five convolutional neural networks (CNNs), to classify retinal fundus photographs into diabetic retinopathy (DR), glaucoma, age-related macular degeneration (AMD) and normal eyes. The CNN architecture was based on the InceptionV3 model, and initial weights were pretrained on the ImageNet dataset. We used 43 055 fundus images from 12 public datasets. Five trained ensembles were then tested on an ‘unseen’ set of 100 images. Seven board-certified ophthalmologists were asked to classify these test images. Results Board-certified ophthalmologists achieved a mean accuracy of 72.7% over all classes, while the DCE achieved a mean accuracy of 79.2% (p=0.03). The DCE had a statistically significant higher mean F1-score for DR classification compared with the ophthalmologists (76.8% vs 57.5%; p=0.01) and greater but statistically non-significant mean F1-scores for glaucoma (83.9% vs 75.7%; p=0.10), AMD (85.9% vs 85.2%; p=0.69) and normal eyes (73.0% vs 70.5%; p=0.39). The DCE had a greater mean agreement between accuracy and confident of 81.6% vs 70.3% (p<0.001). Discussion We developed a deep learning model and found that it could more accurately and reliably classify four categories of fundus images compared with board-certified ophthalmologists. This work provides proof-of-principle that an algorithm is capable of accurate and reliable recognition of multiple retinal diseases using only fundus photographs.