International Classification System for Ocular Complications of Anti-Vascular Endothelial Growth Factor Agents in Clinical Trials.

OBJECTIVE: Complications associated with intravitreal anti-vascular endothelial growth factor (VEGF) therapies are inconsistently reported in the literature, thus limiting an accurate evaluation and comparison of safety between studies. This study aimed to develop a standardized classification system for anti-VEGF ocular complications using the Delphi consensus process. DESIGN: Systematic review and Delphi consensus process. PARTICIPANTS: 25 international retinal specialists participated in the Delphi consensus survey. METHODS: A systematic literature search was conducted to identify complications of intravitreal anti-VEGF agent administration based on randomized controlled trials (RCTs) of anti-VEGF therapy. A comprehensive list of complications was derived from these studies, and this list was subjected to iterative Delphi consensus surveys involving international retinal specialists that voted on inclusion, exclusion, rephrasing, and addition of complications. As well, surveys determined specifiers for the selected complications. This iterative process helped refine the final classification system. MAIN OUTCOME MEASURES: The proportion of retinal specialists who choose to include or exclude complications associated with anti-VEGF administration. RESULTS: After screening 18,229 articles, 130 complications were initially categorized from 145 included RCTs. Participant consensus via the Delphi method resulted in the inclusion of 91 (70%) complications after three rounds. After incorporating further modifications made based on participant suggestions, such as rewording certain phrases and combining similar terms, 24 redundant complications were removed, leaving a total of 67 (52%) complications in the final list. A total of 14 (11%) complications met exclusion thresholds and were eliminated by participants across both rounds. All other remaining complications not meeting inclusion or exclusion thresholds were also excluded from the final classification system after the Delphi process terminated. In addition, 47 out of 75 (63%) proposed complication specifiers were included based on participant agreement. CONCLUSION: Using the Delphi consensus process, a comprehensive, standardized classification system consisting of 67 ocular complications and 47 unique specifiers was established for intravitreal anti-VEGF agents in clinical trials. The adoption of this system in future trials could improve consistency and quality of adverse event reporting, potentially facilitating more accurate risk-benefit analyses.

OCTess: AN OPTICAL CHARACTER RECOGNITION ALGORITHM FOR AUTOMATED DATA EXTRACTION OF SPECTRAL DOMAIN OPTICAL COHERENCE TOMOGRAPHY REPORTS.

PURPOSE: Manual extraction of spectral domain optical coherence tomography (SD-OCT) reports is time and resource intensive. This study aimed to develop an optical character recognition (OCR) algorithm for automated data extraction from Cirrus SD-OCT macular cube reports. METHODS: SD-OCT monocular macular cube reports (n = 675) were randomly selected from a single-center database of patients from 2020 to 2023. Image processing and bounding box operations were performed, and Tesseract (an OCR library) was used to develop the algorithm, OCTess. The algorithm was validated using a separate test data set. RESULTS: The long short-term memory deep learning version of Tesseract achieved the best performance. After reverifying all discrepancies between human and algorithmic data extractions, OCTess achieved accuracies of 100.00% and 99.98% in the training (n = 125) and testing (n = 550) datasets, while the human error rate was 1.11% (98.89% accuracy) and 0.49% (99.51% accuracy) in each, respectively. OCTess extracted data in 3.1 seconds, compared with 94.3 seconds per report for human evaluators. CONCLUSION: We developed an OCR and machine learning algorithm that extracted SD-OCT data with near-perfect accuracy, outperforming humans in both accuracy and efficiency. This algorithm can be used for efficient construction of large-scale SD-OCT data sets for researchers and clinicians.

Exploring the potential utility of AI large language models for medical ethics: an expert panel evaluation of GPT-4.

Integrating large language models (LLMs) like GPT-4 into medical ethics is a novel concept, and understanding the effectiveness of these models in aiding ethicists with decision-making can have significant implications for the healthcare sector. Thus, the objective of this study was to evaluate the performance of GPT-4 in responding to complex medical ethical vignettes and to gauge its utility and limitations for aiding medical ethicists. Using a mixed-methods, cross-sectional survey approach, a panel of six ethicists assessed LLM-generated responses to eight ethical vignettes.The main outcomes measured were relevance, reasoning, depth, technical and non-technical clarity, as well as acceptability of GPT-4's responses. The readability of the responses was also assessed. Of the six metrics evaluating the effectiveness of GPT-4's responses, the overall mean score was 4.1/5. GPT-4 was rated highest in providing technical (4.7/5) and non-technical clarity (4.4/5), whereas the lowest rated metrics were depth (3.8/5) and acceptability (3.8/5). There was poor-to-moderate inter-rater reliability characterised by an intraclass coefficient of 0.54 (95% CI: 0.30 to 0.71). Based on panellist feedback, GPT-4 was able to identify and articulate key ethical issues but struggled to appreciate the nuanced aspects of ethical dilemmas and misapplied certain moral principles.This study reveals limitations in the ability of GPT-4 to appreciate the depth and nuanced acceptability of real-world ethical dilemmas, particularly those that require a thorough understanding of relational complexities and context-specific values. Ongoing evaluation of LLM capabilities within medical ethics remains paramount, and further refinement is needed before it can be used effectively in clinical settings.

ASSOCIATION OF INTRAVENOUS FLUORESCEIN ANGIOGRAPHY AND ADAPTIVE OPTICS IMAGING IN DIABETIC RETINOPATHY: A Prospective Case Series.

PURPOSE: To our knowledge, we present the first case series investigating the relationship between adaptive optics (AO) imaging and intravenous fluorescein angiography (IVFA) parameters in patients with diabetic retinopathy. METHODS: Consecutive patients with diabetic retinopathy older than age 18 years presenting to a single center in Toronto, Canada, from 2020 to 2021 were recruited. Adaptive optics was performed with the RTX1 camera (Imagine Eyes, Orsay, France) at retinal eccentricities of 2° and 4°. Intravenous fluorescein angiography was assessed with the artificial intelligence-based RETICAD system to extract blood flow, perfusion, and blood-retinal-barrier (BRB) permeability at the same retinal locations. Correlations between AO and IVFA parameters were calculated using Pearson's correlation coefficient. RESULTS: Across nine cases, a significant positive correlation existed between photoreceptor spacing on AO and BRB permeability (r = 0.303, P = 0.027), as well as perfusion (r = 0.272, P = 0.049) on IVFA. When stratified by location, a significant positive correlation between photoreceptor dispersion and both BRB permeability and perfusion (r = 0.770, P = 0.043; r = 0.846, P = 0.034, respectively) was observed. Cone density was also negatively correlated with BRB permeability (r = -0.819, P = 0.046). CONCLUSION: Photoreceptor spacing on AO was significantly correlated with BRB permeability and perfusion on IVFA in patients with diabetic retinopathy. Future studies with larger sample sizes are needed to understand the relationship between AO and IVFA parameters in diverse patient populations.

ADAPTIVE OPTICS IMAGING IN DIABETIC RETINOPATHY: A Prospective Cohort Study.

PURPOSE: To investigate the correlation between diabetic retinopathy (DR) severity and microscopic retinal and vascular alterations using adaptive optics imaging. METHODS: In this single-center, prospective cohort study, adult participants with healthy eyes or DR underwent adaptive optics imaging. Participants were classified into control/mild nonproliferative DR, moderate/severe nonproliferative DR, and proliferative DR. Adaptive optics imaging using the RTX1 camera was obtained from 48 participants (87 eyes) for photoreceptor data and from 36 participants (62 eyes) for vascular data. RESULTS: Photoreceptor parameters significantly differed between DR groups at 2° and 4° of retinal eccentricity. Wall-to-lumen ratio varied significantly at 2° eccentricity, while other vascular parameters remained nonsignificant. Cone density and dispersion were the strongest predictors for DR severity ( P < 0.001) in multivariable generalized estimating equation modeling, while other vascular parameters remained nonsignificant between DR severity groups. All photoreceptor parameters showed significant correlations with visual acuity overall and across most DR severity groups. CONCLUSION: To date, this is one of the largest studies evaluating the use of adaptive optics imaging in DR. Adaptive optics imaging was demonstrated to differentiate between various levels of disease severity in DR. These results support the potential role in diagnostic and therapeutic microstructural evaluation in research and clinical practice.

Evaluating ChatGPT on Orbital and Oculofacial Disorders: Accuracy and Readability Insights.

PURPOSE: To assess the accuracy and readability of responses generated by the artificial intelligence model, ChatGPT (version 4.0), to questions related to 10 essential domains of orbital and oculofacial disease. METHODS: A set of 100 questions related to the diagnosis, treatment, and interpretation of orbital and oculofacial diseases was posed to ChatGPT 4.0. Responses were evaluated by a panel of 7 experts based on appropriateness and accuracy, with performance scores measured on a 7-item Likert scale. Inter-rater reliability was determined via the intraclass correlation coefficient. RESULTS: The artificial intelligence model demonstrated accurate and consistent performance across all 10 domains of orbital and oculofacial disease, with an average appropriateness score of 5.3/6.0 ("mostly appropriate" to "completely appropriate"). Domains of cavernous sinus fistula, retrobulbar hemorrhage, and blepharospasm had the highest domain scores (average scores of 5.5 to 5.6), while the proptosis domain had the lowest (average score of 5.0/6.0). The intraclass correlation coefficient was 0.64 (95% CI: 0.52 to 0.74), reflecting moderate inter-rater reliability. The responses exhibited a high reading-level complexity, representing the comprehension levels of a college or graduate education. CONCLUSIONS: This study demonstrates the potential of ChatGPT 4.0 to provide accurate information in the field of ophthalmology, specifically orbital and oculofacial disease. However, challenges remain in ensuring accurate and comprehensive responses across all disease domains. Future improvements should focus on refining the model's correctness and eventually expanding the scope to visual data interpretation. Our results highlight the vast potential for artificial intelligence in educational and clinical ophthalmology contexts.

Profiling medical specialties and informing aspiring physicians: a data-driven approach.

A detailed, unbiased perspective of the inter-relations among medical fields could help students make informed decisions on their future career plans. Using a data-driven approach, the inter-relations among different medical fields were decomposed and clustered based on the similarity of their working environments.Publicly available, aggregate databases were merged into a single rich dataset containing demographic, working environment and remuneration information for physicians across Canada. These data were collected from the Canadian Institute for Health Information, the Canadian Medical Association, and the Institute for Clinical Evaluative Sciences, primarily from 2018 to 2019. The merged dataset includes 25 unique medical specialties, each with 36 indicator variables. Latent Profile Analysis (LPA) was used to group specialties into distinct clusters based on relatedness.The 25 medical specialties were decomposed into seven clusters (latent variables) that were chosen based on the Bayesian Information Criterion. The Kruskal-Wallis test identified eight indicator variables that significantly differed between the seven profiles. These variables included income, work settings and payment styles. Variables that did not significantly vary between profiles included demographics, professional satisfaction, and work-life balance satisfaction.The 25 analyzed medical specialties were grouped in an unsupervised manner into seven profiles via LPA. These profiles correspond to expected and meaningful groups of specialties that share a common theme and set of indicator variables (e.g. procedurally-focused, clinic-based practice). These profiles can help aspiring physicians narrow down and guide specialty choice.

The Isabel Differential Diagnosis Generator for Orbital Diagnosis.

PURPOSE: The Isabel differential diagnosis generator is one of the most widely known electronic diagnosis decision support tools. The authors prospectively evaluated the utility of Isabel for orbital disease differential diagnosis. METHODS: The terms "proptosis," "lid retraction," "orbit inflammation," "orbit tumour," "orbit tumor, infiltrative" and "orbital tumor, well-circumscribed" were separately input into Isabel and the results were tabulated. Then the clinical details (patient age, gender, signs, symptoms, and imaging findings) of 25 orbital cases from a textbook of orbital surgery were entered into Isabel. The top 10 differential diagnoses generated by Isabel were compared with the correct diagnosis. RESULTS: Isabel identified hyperthyroidism and Graves ophthalmopathy as the leading causes of lid retraction, but many common causes of proptosis and orbital tumors were not correctly elucidated. Of the textbook cases, Isabel correctly identified 4/25 (16%) of orbital cases as one of its top 10 differential diagnoses, and the median rank of the correct diagnosis was 6/10. Thirty-two percent of the output diagnoses were unlikely to cause orbital disease. CONCLUSION: Isabel is currently of limited value in the mainstream orbital differential diagnosis. The incorporation of anatomic localizations and imaging findings may help increase the accuracy of orbital diagnosis.

Inflammatory Biomarkers and Intracranial Hemorrhage after Endovascular Thrombectomy.

BACKGROUND: Intracranial hemorrhage after endovascular thrombectomy is associated with poorer prognosis compared with those who do not develop the complication. Our study aims to determine predictors of post-EVT hemorrhage - more specifically, inflammatory biomarkers present in baseline serology. METHODS: We performed a retrospective review of consecutive patients treated with EVT for acute large vessel ischemic stroke. The primary outcome of the study is the presence of ICH on the post-EVT scan. We used four definitions: the SITS-MOST criteria, the NINDS criteria, asymptomatic hemorrhage, and overall hemorrhage. We identified nonredundant predictors of outcome using backward elimination based on Akaike Information Criteria. We then assessed prediction accuracy using area under the receiver operating curve. Then we implemented variable importance ranking from logistic regression models using the drop in Naegelkerke R2 with the exclusion of each predictor. RESULTS: Our study demonstrates a 6.3% SITS (16/252) and 10.0% NINDS (25/252) sICH rate, as well as a 19.4% asymptomatic (49/252) and 29.4% (74/252) overall hemorrhage rate. Serologic markers that demonstrated association with post-EVT hemorrhage were: low lymphocyte count (SITS), high neutrophil count (NINDS, overall hemorrhage), low platelet to lymphocyte ratio (NINDS), and low total WBC (NINDS, asymptomatic hemorrhage). CONCLUSION: Higher neutrophil counts, low WBC counts, low lymphocyte counts, and low platelet to lymphoycyte ratio were baseline serology biomarkers that were associated with post-EVT hemorrhage. Our findings, particularly the association of diabetes mellitus and high neutrophil, support experimental data on the role of thromboinflammation in hemorrhagic transformation of large vessel occlusions.

Gender-based differences in physician payments within the fee-for-service system in Ontario: a retrospective, cross-sectional study.

BACKGROUND: Differences in physician income by gender have been described in numerous jurisdictions, but few studies have looked at a Canadian cohort with adjustment for confounders. In this study, we aimed to understand differences in fee-for-service payments to men and women physicians in Ontario. METHODS: We conducted a cross-sectional analysis of all Ontario physicians who submitted claims to the Ontario Health Insurance Plan (OHIP) in 2017. For each physician, we gathered demographic information from the College of Physicians and Surgeons of Ontario registry. We compared differences in physician claims between men and women in the entire cohort and within each specialty using multivariable linear regressions, controlling for length of practice, specialty and practice location. RESULTS: We identified a cohort of 30 167 physicians who submitted claims to OHIP in 2017, including 17 992 men and 12 175 women. When controlling for confounding variables in a linear mixed-effects regression model, annual physician claims were $93 930 (95% confidence interval $88 434 to $99 431) higher for men than for women. Women claimed 74% as much as men when adjusting for covariates. This discrepancy was present in nearly all specialty categories. Men claimed more than women throughout their careers, with the greatest gap 10-15 years into practice. INTERPRETATION: We found a gender gap in fee-for-service claims in Ontario, with women claiming less than men overall and in nearly every specialty. Further work is required to understand the root causes of the gender pay gap.

Data Driven Analysis Reveals Shared Transcriptome Response, Immune Cell Composition, and Distinct Mortality Rates Across Differing Etiologies of Critical Illness.

OBJECTIVES: Sepsis and trauma are common health problems and provide great challenges in critical care. Diverse patient responses to these conditions further complicate patient management and outcome prediction. Whole blood transcriptomics provides a unique opportunity to follow the molecular response in the critically ill. Prior results show robust and diverse genomic signal in the acute phase and others have found shared biological mechanisms across divergent disease etiologies. We hypothesize that selected transcriptomics responses, particularly immune mechanisms are shared across disease etiologies. We further hypothesize that these processes may identify homogenous patient subgroups with shared clinical course in critical illness deciphering disease heterogeneity. These processes may serve as universal markers for predicting a complicated clinical course and/or risk of a poor outcome. DESIGN: We present a system level, data driven, genome-wide analysis of whole blood gene expression for a total of 382 patients suffering from either abdominal sepsis (49), pulmonary sepsis (107) or trauma (158) and compare these to gene expression in healthy controls (68). PATIENTS AND SETTING: We relied on available open genetic data from gene expression omnibus for patients diagnosed with abdominal sepsis, community-acquired pneumonia, or trauma which also included healthy control patients. MEASUREMENTS AND MAIN RESULTS: Our results confirm that immune processes are shared across disease etiologies in critical illnesses. We identify two consistent and distinct patient subgroups through deconvolution of serum transcriptomics: 1) increased neutrophils and naïve CD4 cell fractions and 2) suppressed neutrophil fraction. Furthermore, we found immune and inflammatory processes were downregulated in subgroup 2, a configuration previously shown to be more susceptible to multiple organ failure. Correspondingly, this subgroup had significantly higher mortality rates in all three etiologies of illness (0% vs 6.1%, p = 3.1 × 10 for trauma; 15.0% vs 25.4%, p = 4.4 × 10 for community-acquired pneumonia, and 7.1% vs 20.0%, p = 3.4 × 10 for abdominal sepsis). CONCLUSIONS: We identify two consistent subgroups of critical illness based on serum transcriptomics and derived immune cell fractions, with significantly different survival rates. This may serve as a universal predictor of complicated clinical course or treatment response and, importantly, may identify opportunities for subgroup-specific immunomodulatory intervention.

Dysphotopsia and location of laser iridotomy: a systematic review.

This systematic review aimed to clarify the relationship between the location of laser peripheral iridotomy (LPI), a common procedure to prevent or treat angle-closure glaucoma, and the incidence of post-procedure visual disturbances known as dysphotopsias. Understanding this relationship is crucial due to the high frequency of LPIs performed and the significant impact dysphotopsia can have on vision and quality of life. Articles investigating the relationship between LPI location and dysphotopsia in at least five patients were identified via a literature search of OVID MEDLINE (1946-November 19, 2022) and EMBASE (1946-November 19, 2022). Non-comparative and non-English studies were excluded. Studies did not require a control group to be included. The Grading of Recommendations, Assessment, Development and Evaluation (GRADE) system and Cochrane Risk of Bias 2 (RoB2) tool were used to appraise included studies. Our review included three studies encompassing 1756 eyes from 878 patients. The location of LPI was grouped into superior (604 patients, 889 eyes), inferior (150 patients, 150 eyes), and nasal/temporal (443 patients, 717 eyes). The analysis showed no significant difference in the incidence of any new dysphotopsia types among the location groups post-LPI. Overall, the incidence of lines, ghost images, and blurring significantly increased after LPI, while halos and glare significantly decreased. In conclusion, the current literature suggests that the location of LPI has no significant relationship to the types and rates of dysphotopsia experienced thereafter. While there is a 2-3% risk of linear dysphotopsia after LPI regardless of location, LPI may also resolve pre-existing halos and glare.

Early Discharge after Uncomplicated Elective Colectomy and Risk of Postdischarge Complication.

BACKGROUND: This was a retrospective cohort study of adult patients undergoing uncomplicated elective colectomy using the NSQIP database from January 2012 to December 2019. A colectomy is deemed uncomplicated if there are no complications reported during the hospitalization. The objective of this study was to examine the association between discharge timing and postdischarge complications in patients who undergo uncomplicated elective colectomy. STUDY DESIGN: Patients were stratified into an early discharge group if their length of postoperative hospitalization was ≤3 days for laparoscopic or robotic approaches, or ≤5 days for the open approach, and otherwise into delayed discharge groups. The association between early discharge and any postdischarge complication was examined using unadjusted logistic regression after propensity score matching between early and delayed discharge groups. RESULTS: Of the 113,940 patients included, 77,979, 15,877, and 20,084 patients underwent uncomplicated laparoscopic, robotic, and open colectomy, respectively. After propensity score matching, the odds of a postdischarge complication were lower for the early discharge group in laparoscopic (odds ratio 0.73, 95% CI 0.68 to 0.79) and robotic (odds ratio 0.63, 95% CI 0.52 to 0.76) approaches, and not different in the open approach (odds ratio 1.02, 95% CI 0.91 to 1.15). There were no clinically meaningful differences in the risk of return to the operating room for all surgical approaches. CONCLUSIONS: Early discharge after uncomplicated colectomy appears to be safe and is associated with lower odds of postdischarge complications in minimally invasive approaches. Our findings suggest that surgical teams practice sound clinical judgments on selecting patients who benefit from early discharge.

Does routine upper gastrointestinal swallow study after metabolic and bariatric surgery lead to earlier diagnosis of leak?

BACKGROUND: It is unclear whether routine upper gastrointestinal swallow study (SS) in the immediate postoperative period is associated with earlier diagnosis of gastrointestinal leak after bariatric surgery. OBJECTIVE: To investigate the relationship between routine SS and time to diagnosis of postoperative gastrointestinal leak. SETTING: MBSAQIP-accredited hospitals in the United States and Canada. METHODS: We conducted an observational cohort study of adults who underwent laparoscopic primary Roux-en-Y gastric bypass (RYGB) (n = 82,510) and sleeve gastrectomy (SG) (n = 283,520) using the MBSAQIP 2015-2019 database. Propensity scores were used to match patient cohorts who underwent routine versus no routine SS. Primary outcome was time to diagnosis of leak. Median days to diagnosis of leak were compared. The Nelson-Aalen estimator was used to determine the cumulative hazards of leak. RESULTS: In our study, 36,280 (23%) RYGB and 135,335 (33%) SG patients received routine SS. Routine SS was not associated with earlier diagnosis of leak (RYGB routine SS median 7 [IQR 3-12] days v. no routine SS 6 [2-11] days, P = .9; SG routine SS 15 [9-22] days v. no routine SS 14 [8-21] days, P = .06) or lower risk of developing leak (RYGB HR 1.0, 95%-CI .8-1.2; SG HR 1.1, 95%-CI 1.0-1.4). More routine SS patients had a length of stay 2 days or greater (RYGB 78.3% v. 61.1%; SG 48.6% v. 40.3%). CONCLUSIONS: Routine SS was not associated with earlier diagnosis of leaks compared to the absence of routine SS. Surgeons should consider abandoning the practice of routine SS for the purpose of obtaining earlier diagnosis of postoperative leaks.

Pars plana vitrectomy versus intravitreal antibiotics for endophthalmitis management following intravitreal anti-VEGF agents: A meta-analysis.

In this meta-analysis, we aim to compare the efficacy and safety of pars plana vitrectomy (PPV) versus tap-and-inject (TAI) of intravitreal antibiotics for the management of endophthalmitis secondary to intravitreal injections of anti-vascular endothelial growth factor (VEGF) agents. A systematic literature search was conducted on Ovid MEDLINE, EMBASE, and Cochrane Central (January 2005-October 2022). The primary analysis compared initial PPV versus TAI and the secondary analysis examined the efficacy and safety of TAI alone compared to TAI followed by PPV. The quality of non-randomized observational studies was assessed using the Newcastle-Ottawa Scale. The quality of the evidence was assessed for each outcome. A random effects meta-analysis was performed. Weighted mean differences (WMDs) with 95% confidence intervals were reported. Of the 7474 screened studies, nine studies reporting on 153 eyes were included. The change in mean best corrected visual acuity (BCVA) between endophthalmitis presentation and last follow-up did not significantly differ between the initial TAI versus PPV groups (WMD = 0.05 units; 95% CI -0.12 to 0.22; p = 0.59; heterogeneity p = 0.41). The difference in pre- to post-treatment mean BCVA did not significantly differ between eyes that received TAI alone or TAI followed by PPV (WMD = 0.04 units; 95% CI -0.42 to 0.51; p = 0.85; heterogeneity p = 0.74). While the meta-analysis revealed no significant difference in the change in BCVA between PPV and TAI to treat endophthalmitis secondary to anti-VEGF agents, the quality of evidence was low with potential for confounding and selection bias. Further well-designed studies in this setting are needed.

Systemic Arterial and Venous Thrombotic Events Associated With Anti-Vascular Endothelial Growth Factor Injections: A Meta-Analysis.

PURPOSE: To compare the risk of systemic arteriovenous thrombotic events between intravitreal anti-vascular endothelial growth factor (anti-VEGF) and sham injections. DESIGN: Random-effects meta-analysis. METHODS: A systematic search was performed on OVID MEDLINE, Embase, and Cochrane Library from January 2005 to August 2023. Our inclusion criteria were randomized controlled trials (RCTs) reporting on systemic arteriovenous events for standard dose intravitreal anti-VEGF agents for any indication. RESULTS: A total of 20 RCTs reporting on 12,833 eyes were included. There was no significant difference in the risk of any thrombotic event between bevacizumab 1.25 mg and ranibizumab 0.5 mg (Risk ratio (RR) = 0.96, 95% CI = 0.52-1.75, P = .89). There was no significant difference between bevacizumab and ranibizumab when restricting to arterial thrombotic events (RR= 0.88, 95% CI = 0.60-1.30, P = .53) or venous thrombotic events (RR = 1.99, 95% CI =86 0.68-5.82], P = .21). The risk of arterial thrombotic events was similar between aflibercept and bevacizumab (RR = 1.11, 95% CI = 0.60-2.07, P = .74), between aflibercept and ranibizumab (RR= 0.77, 95% CI = 0.49-1.21, P = .26), between brolucizumab and aflibercept (RR= 0.67, 95% CI = 0.32-1.38, P = .27), and between aflibercept and faricimab (RR = 0.96, 95% CI = 0.43-2.17, P = .93). Compared to sham, neither dose of ranibizumab (0.5 mg or 0.3 mg) showed a higher risk of arterial thrombotic events. CONCLUSIONS: There was a similar risk of systemic arteriovenous thrombotic adverse events between anti-VEGF agents and between ranibizumab and sham injections.