Search

Importance: Machine learning (ML) algorithms have the potential to identify eyes with early diabetic retinopathy (DR) at increased risk for disease progression., Objective: To create and validate automated ML models (autoML) for DR progression from ultra-widefield (UWF) retinal images., Design, Setting and Participants: Deidentified UWF images with mild or moderate nonproliferative DR (NPDR) with 3 years of longitudinal follow-up retinal imaging or evidence of progression within 3 years were used to develop automated ML models for predicting DR progression in UWF images. All images were collected from a tertiary diabetes-specific medical center retinal image dataset. Data were collected from July to September 2022., Exposure: Automated ML models were generated from baseline on-axis 200° UWF retinal images. Baseline retinal images were labeled for progression based on centralized reading center evaluation of baseline and follow-up images according to the clinical Early Treatment Diabetic Retinopathy Study severity scale. Images for model development were split 8-1-1 for training, optimization, and testing to detect 1 or more steps of DR progression. Validation was performed using a 328-image set from the same patient population not used in model development., Main Outcomes and Measures: Area under the precision-recall curve (AUPRC), sensitivity, specificity, and accuracy., Results: A total of 1179 deidentified UWF images with mild (380 [32.2%]) or moderate (799 [67.8%]) NPDR were included. DR progression was present in half of the training set (590 of 1179 [50.0%]). The model's AUPRC was 0.717 for baseline mild NPDR and 0.863 for moderate NPDR. On the validation set for eyes with mild NPDR, sensitivity was 0.72 (95% CI, 0.57-0.83), specificity was 0.63 (95% CI, 0.57-0.69), prevalence was 0.15 (95% CI, 0.12-0.20), and accuracy was 64.3%; for eyes with moderate NPDR, sensitivity was 0.80 (95% CI, 0.70-0.87), specificity was 0.72 (95% CI, 0.66-0.76), prevalence was 0.22 (95% CI, 0.19-0.27), and accuracy was 73.8%. In the validation set, 6 of 9 eyes (75%) with mild NPDR and 35 of 41 eyes (85%) with moderate NPDR progressed 2 steps or more were identified. All 4 eyes with mild NPDR that progressed within 6 months and 1 year were identified, and 8 of 9 (89%) and 17 of 20 (85%) with moderate NPDR that progressed within 6 months and 1 year, respectively, were identified., Conclusions and Relevance: This study demonstrates the accuracy and feasibility of automated ML models for identifying DR progression developed using UWF images, especially for prediction of 2-step or greater DR progression within 1 year. Potentially, the use of ML algorithms may refine the risk of disease progression and identify those at highest short-term risk, thus reducing costs and improving vision-related outcomes.

Background: Diabetic macular oedema (DMO) is a common complication of diabetic retinopathy. Antiangiogenic therapy with anti-vascular endothelial growth factor (anti-VEGF) can reduce oedema, improve vision, and prevent further visual loss. These drugs have replaced laser photocoagulation as the standard of care for people with DMO. In the previous update of this review, we found moderate-quality evidence that, at 12 months, aflibercept was slightly more effective than ranibizumab and bevacizumab for improving vision in people with DMO, although the difference may have been clinically insignificant (less than 0.1 logarithm of the minimum angle of resolution (logMAR), or five Early Treatment Diabetic Retinopathy Study (ETDRS) letters, or one ETDRS line)., Objectives: The objective of this updated review was to compare the effectiveness and safety of the different anti-VEGF drugs in RCTs at longer followup (24 months)., Search Methods: We searched various electronic databases on 8 July 2022., Selection Criteria: We included randomised controlled trials (RCTs) that compared any anti-angiogenic drug with an anti-VEGF mechanism of action versus another anti-VEGF drug, another treatment, sham, or no treatment in people with DMO., Data Collection and Analysis: We used standard Cochrane methods for pairwise meta-analysis and we augmented this evidence using network meta-analysis (NMA) methods. We used the Stata 'network' meta-analysis package for all analyses. We used the CINeMA (Confidence in Network Meta-Analysis) web application to grade the certainty of the evidence., Main Results: We included 23 studies (13 with industry funding) that enrolled 3513 people with DMO (median central retinal thickness (CRT) 460 microns, interquartile range (IQR) 424 to 482) and moderate vision loss (median best-corrected visual acuity (BCVA) 0.48 logMAR, IQR 0.42 to 0.55. One study that investigated ranibizumab versus sham and one study that mainly enrolled people with subclinical DMO and normal BCVA were not suitable for inclusion in the efficacy NMA. Consistent with the previous update of this review, we used ranibizumab as the reference drug for efficacy, and control (including laser, observation, and sham) as the reference for systemic safety. Eight trials provided data on the primary outcome (change in BCVA at 24 months, in logMAR: lower is better). We found no evidence of a difference between the following interventions and ranibizumab alone: aflibercept (mean difference (MD) -0.05 logMAR, 95% confidence interval (CI) -0.12 to 0.02; moderate certainty); bevacizumab (MD -0.01 logMAR, 95% CI -0.13 to 0.10; low certainty), brolucizumab (MD 0.00 logMAR, 95% CI -0.08 to 0.07; low certainty), ranibizumab plus deferred laser (MD 0.00 logMAR, 95% CI -0.11 to 0.10; low certainty), and ranibizumab plus prompt laser (MD 0.03 logMAR, 95% CI -0.04 to 0.09; very low certainty). We also analysed BCVA change at 12 months, finding moderate-certainty evidence of increased efficacy with brolucizumab (MD -0.07 logMAR, 95%CI -0.10 to -0.03 logMAR), faricimab (MD -0.08 logMAR, 95% CI -0.12 to -0.05), and aflibercept (MD -0.07 logMAR, 95 % CI -0.10 to -0.04) compared to ranibizumab alone, but the difference could be clinically insignificant. Compared to ranibizumab alone, NMA of six trials showed no evidence of a difference with aflibercept (moderate certainty), bevacizumab (low certainty), or ranibizumab with prompt (very low certainty) or deferred laser (low certainty) regarding improvement by three or more ETDRS lines at 24 months. There was moderate-certainty evidence of greater CRT reduction at 24 months with brolucizumab (MD -23 microns, 95% CI -65 to -1 9) and aflibercept (MD -26 microns, 95% CI -53 to 0.9) compared to ranibizumab. There was moderate-certainty evidence of lesser CRT reduction with bevacizumab (MD 28 microns, 95% CI 0 to 56), ranibizumab plus deferred laser (MD 63 microns, 95% CI 18 to 109), and ranibizumab plus prompt laser (MD 72 microns, 95% CI 25 to 119) compared with ranibizumab alone. Regarding all-cause mortality at the longest available follow-up (20 trials), we found no evidence of increased risk of death for any drug compared to control, although effects were in the direction of an increase, and clinically relevant increases could not be ruled out. The certainty of this evidence was low for bevacizumab (risk ratio (RR) 2.10, 95% CI 0.75 to 5.88), brolucizumab (RR 2.92, 95% CI 0.68 to 12.58), faricimab (RR 1.91, 95% CI 0.45 to 8.00), ranibizumab (RR 1.26, 95% CI 0.68 to 2.34), and very low for conbercept (RR 0.33, 95% CI 0.01 to 8.81) and aflibercept (RR 1.48, 95% CI 0.79 to 2.77). Estimates for Antiplatelet Trialists Collaboration arterial thromboembolic events at 24 months did not suggest an increase with any drug compared to control, but the NMA was overall incoherent and the evidence was of low or very low certainty. Ocular adverse events were rare and poorly reported and could not be assessed in NMAs., Authors' Conclusions: There is limited evidence of the comparative efficacy and safety of anti-VEGF drugs beyond one year of follow-up. We found no clinically important differences in visual outcomes at 24 months in people with DMO, although there were differences in CRT change. We found no evidence that any drug increases all-cause mortality compared to control, but estimates were very imprecise. Evidence from RCTs may not apply to real-world practice, where people in need of antiangiogenic treatment are often under-treated, and the individuals exposed to these drugs may be less healthy than trial participants., Competing Interests: GV: none known KC: none known EL: none known TP: received honoraria for advisory board and lecture fees from Bayer, Allergan, B-I, Sandoz, Roche, Novartis, Heidelberg, Zeiss, and Optos MP: received payment for participating on the Advisory Board for Allergan, Bayer and Novartis. Jennifer Evans (methods guide, not an author): none known Iris Gordon (search specialist, not an author): none known, (Copyright © 2023 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.)

Background: Diabetes is a major public health concern, with approximately 80% of the burden falling on low- and middle-income countries (LMICs). Diabetic retinopathy (DR) is one of the most common microvascular complications of diabetes, and early detection through diabetic eye screening programmes is essential to prevent visual impairment and blindness. Careful planning at a national level is crucial for effective implementation of such programmes., Methods: A scoping review was conducted, and the protocol was published previously to explain the methods in detail. Data were collected from databases and searches, including grey literature. Furthermore, consultations were conducted with key informants from LMICs., Results: Lower-middle-income countries (29/50, 58%) and upper-middle-income countries (27/59, 45.8%) are making more progress than low-income countries (4/29, 13.8%) in terms of DR policy planning. However, no identified data for published policies have actually implemented national DR policies. Compared to low-income and lower-middle-income countries, upper-middle-income countries are making the most progress in implementing national diabetic eye screening programmes; however, their progress is still slow, with only 5/59 (8.5%) having such programmes., Conclusion: There are significant gaps in the literature, with no data reported for 78/138 (56.5%) LMICs. Further research is clearly needed to support and document DR policy development in LMICs., (© 2022. The Author(s).)

Objectives: To compare the accuracy of trained level 1 diabetic retinopathy (DR) graders (nurses, endocrinologists and one general practitioner), level 2 graders (midlevel ophthalmologists) and level 3 graders (senior ophthalmologists) in Vietnam against a reference standard from the UK and assess the impact of supplementary targeted grader training., Design: Diagnostic test accuracy study., Setting: Secondary care hospitals in Southern Vietnam., Participants: DR training was delivered to Vietnamese graders in February 2018 by National Health Service (NHS) UK graders. Two-field retinal images (412 patient images) were graded by 14 trained graders in Vietnam between August and October 2018 and then regraded retrospectively by an NHS-certified reference standard UK optometrist (phase I). Further DR training based on phase I results was delivered to graders in November 2019. After training, a randomised subset of images from January to October 2020 (115 patient images) was graded by six of the original cohort (phase II). The reference grader regraded all images from phase I and II retrospectively in masked fashion., Primary and Secondary Outcome Measures: Sensitivity was calculated at the two different time points, and χ 2 was used to test significance., Results: In phase I, the sensitivity for detecting any DR for all grader groups in Vietnam was low (41.8-42.2%) and improved in phase II after additional training was delivered (51.3-87.2%). The greatest improvement was seen among level 1 graders (p<0.001), and the lowest improvement was observed among level 3 graders (p=0.326). There was a statistically significant improvement in sensitivity for detecting referable DR and referable diabetic macular oedema between all grader levels. The post-training values ranged from 40.0 to 61.5% (including ungradable images) and 55.6%-90.0% (excluding ungradable images)., Conclusions: This study demonstrates that targeted training interventions can improve accuracy of DR grading. These findings have important implications for improving service delivery in DR screening programmes in low-resource settings., Competing Interests: Competing interests: NC is employed as a Research Director by Orbis International., (© Author(s) (or their employer(s)) 2022. Re-use permitted under CC BY. Published by BMJ.)

Introduction: Globally, diabetic retinopathy (DR) is the leading cause of blindness in working-aged adults. Early detection and treatment of DR is essential for preventing sight loss. Services must be available, accessible and acceptable to patients if we are to ensure they seek such care., Objectives: To understand patients' knowledge and attitudes towards laser versus antivascular endothelial growth factor (VEGF) injections to treat DR in Vietnam, and to identify factors Vietnamese ophthalmologists consider when making treatment decisions., Methods: This is a descriptive qualitative study based on semi-structured interviews with 18 patients (12 from Ho Chi Minh City and 6 from Hanoi) plus individual interviews with 24 ophthalmologists working in eye clinics in these cities. Thematic analysis was used to analyse the data., Results: In total, 10/24 (41.7%) ophthalmologists were female, and their median age was 41 years (range 29-69 years). The median age of patients was 56.5 years (range 28-72 years), and 7/18 (38.9%) were female. Briefly, factors that influence DR treatment decisions for ophthalmologists are medical considerations (ie, severity of disease, benefits and risks), availability (ie, treatment and resources) and patient-related factors (ie, costs and adherence). Patient's perceived barriers and facilitators to treatments were based on patient and family related factors (ie, treatment and transportation costs) and previous treatment experiences (ie, positive and negative). Recommendations by all participants included ensuring that both laser and anti-VEGF injections are widely available across the country and controlling costs for patients and the healthcare system., Conclusions: Reducing DR treatment costs, optimising treatments options, and expanding the network of clinics offering treatment outside metropolitan areas were the main issues raised by participants. These findings can help inform policy changes in Vietnam and may be generalisable to other low-resource settings., Competing Interests: Competing interests: NC is employed as a Research Director by Orbis International., (© Author(s) (or their employer(s)) 2022. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ.)

Background: Diabetic eye screening programmes have been developed worldwide based on evidence that early detection and treatment of diabetic retinopathy are crucial to preventing sight loss. However, little is known about the decision-making processes and training needs of diabetic retinal graders, particularly in low- and middle-income countries., Objectives: To provide data for improving evidence-based diabetic retinopathy training to help novice graders process fundus images more like experts., Subjects/methods: This is a mixed-methods qualitative study conducted in southern Vietnam and Northern Ireland. Novice diabetic retinal graders in Vietnam (n = 18) and expert graders in Northern Ireland (n = 5) were selected through a purposive sampling technique. Data were collected from 21st February to 3rd September 2019. The interviewer used neutral prompts during think-aloud sessions to encourage participants to verbalise their thought processes while grading fundus images from anonymised patients, followed by semi-structured interviews. Thematic framework analysis was used to identify themes, supported by illustrative quotes from interviews. Mann-Whitney U tests were used to compare graders' performance., Results: Expert graders used a more systematic approach when grading images, considered all four images per patient and used available software tools such as red-free filters prior to making a decision on management. The most challenging features for novice graders were intra-retinal microvascular abnormalities and new vessels, which were more accurately identified by experts., Conclusion: Taking more time to grade fundus images and adopting a protocol-driven "checklist" approach may help novice graders to function more like experts., (© 2021. The Author(s).)

Introduction: The diabetes mellitus (DM) epidemic is a major public health concern globally, with the highest-burden in low-income and middle-income countries (LMICs). Diabetic retinopathy (DR) is a microvascular complication of diabetes, and if left untreated can lead to visual impairment and blindness. Epidemiological studies suggest that the incidence of sight-threatening DR is decreasing in high-income countries due to improved treatments and management of DM; however, these trends are not replicated in LMICs. In this paper, we outline a scoping review protocol that aims to identify which LMICs have included DR in their national DM, non-communicable disease or prevention of blindness plans. The scoping review also aims to assess gaps when implementing national DR screening programmes in LMICs., Methods and Analysis: This scoping review will follow the Arksey and O'Malley (2005) methodology and the Preferred Reporting Items for Systematic Review and Meta-Analysis extension for Scoping Review guidelines. A comprehensive search of peer-reviewed and grey literature will be conducted from October 1989 (St. Vincent Declaration) to February 2020. Studies will be identified from electronic databases; Medline, Embase and CENTRAL (Cochrane Library). To identify further relevant articles, a hand search will be conducted using the reference lists of included studies. Two reviewers will independently screen records for relevant data and disagreements about eligibility will be resolved through consensus or arbitration by a third reviewer. A quantitative analysis will be performed to highlight key findings and thematic analysis will be used to identify emerging themes and subthemes from included studies. The key themes will highlight countries progress in terms of national-level DR service planning and screening implementation., Ethics and Dissemination: No ethical approval is required because the scoping review methodology aims to synthesise information from publicly available resources. The results will be disseminated through conference presentations and peer-reviewed publication., Competing Interests: Competing interests: None declared., (© Author(s) (or their employer(s)) 2020. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ.)