Semaglutide and Nonarteritic Anterior Ischemic Optic Neuropathy

Cindy X Cai; Michelle Hribar; Sally Baxter; Kerry Goetz; Swarup S Swaminathan; Alexis Flowers; Eric N Brown; Brian Toy; Benjamin Xu; John Chen; Aiyin Chen; Sophia Wang; Cecilia Lee; Theodore Leng; Joshua R Ehrlich; Andrew Barkmeier; Karen R Armbrust; Michael V Boland; David Dorr; Danielle Boyce; Thamir Alshammari; Joel Swerdel; Marc A Suchard; Martijn Schuemie; Fan Bu; Anthony G Sena; George Hripcsak; Akihiko Nishimura; Paul Nagy; Thomas Falconer; Scott L DuVall; Michael Matheny; Benjamin Viernes; William O’Brien; Linying Zhang; Benjamin Martin; Erik Westlund; Nestoras Mathioudakis; Ruochong Fan; Adam Wilcox; Albert Lai; Jacqueline C Stocking; Sahar Takkouche; Lok Hin Lee; Yangyiran Xie; Izabelle Humes; David B McCoy; Mohammad Adibuzzaman; Raymond G Areaux, Jr; William Rojas-Carabali; James Brash; David A Lee; Nicole G Weiskopf; Louise Mawn; Rupesh Agrawal; Hannah Morgan-Cooper; Priya Desai; Patrick B Ryan

doi:10.1001/jamaophthalmol.2024.6555

. 2025 Feb 20;143(4):304–314. doi: 10.1001/jamaophthalmol.2024.6555

Semaglutide and Nonarteritic Anterior Ischemic Optic Neuropathy

Cindy X Cai ^1,^2,^✉, Michelle Hribar ^3,⁴, Sally Baxter ^5,⁶, Kerry Goetz ³, Swarup S Swaminathan ⁷, Alexis Flowers ^8,⁹, Eric N Brown ⁹, Brian Toy ¹⁰, Benjamin Xu ¹⁰, John Chen ^11,¹², Aiyin Chen ⁴, Sophia Wang ¹³, Cecilia Lee ^14,¹⁵, Theodore Leng ¹⁶, Joshua R Ehrlich ^17,¹⁸, Andrew Barkmeier ¹¹, Karen R Armbrust ¹⁹, Michael V Boland ²⁰, David Dorr ²¹, Danielle Boyce ^22,²³, Thamir Alshammari ^24,²⁵, Joel Swerdel ²⁶, Marc A Suchard ^27,²⁸, Martijn Schuemie ^26,²⁷, Fan Bu ²⁹, Anthony G Sena ^26,³⁰, George Hripcsak ³¹, Akihiko Nishimura ³², Paul Nagy ³³, Thomas Falconer ³¹, Scott L DuVall ^28,³⁴, Michael Matheny ³⁵, Benjamin Viernes ²⁸, William O’Brien ²⁸, Linying Zhang ³⁶, Benjamin Martin ², Erik Westlund ³², Nestoras Mathioudakis ³⁷, Ruochong Fan ³⁶, Adam Wilcox ³⁸, Albert Lai ³⁸, Jacqueline C Stocking ³⁹, Sahar Takkouche ⁴⁰, Lok Hin Lee ⁴¹, Yangyiran Xie ⁴¹, Izabelle Humes ⁴², David B McCoy ⁴², Mohammad Adibuzzaman ⁴², Raymond G Areaux Jr ¹⁹, William Rojas-Carabali ⁴³, James Brash ⁴⁴, David A Lee ⁴⁵, Nicole G Weiskopf ⁴⁶, Louise Mawn ⁴⁷, Rupesh Agrawal ⁴⁸, Hannah Morgan-Cooper ⁴⁹, Priya Desai ⁴⁹, Patrick B Ryan ^50,⁵¹

¹Wilmer Eye Institute, Johns Hopkins School of Medicine, Baltimore, Maryland

²Biomedical Informatics and Data Science, Division of General Internal Medicine, Department of Medicine, Johns Hopkins University School of Medicine, Baltimore, Maryland

³National Eye Institute, National Institutes of Health, Bethesda, Maryland

⁴Casey Eye Institute, Department of Medical Informatics and Clinical Epidemiology, Oregon Health & Science University, Portland

⁵Viterbi Family Department of Ophthalmology and Shiley Eye Institute, University of California San Diego, La Jolla

⁶Division of Biomedical Informatics, Department of Medicine, University of California San Diego, La Jolla

⁷Department of Ophthalmology, Bascom Palmer Eye Institute, University of Miami Miller School of Medicine, Miami, Florida

⁸Vanderbilt Eye Institute, Department of Neurology, Vanderbilt University Medical Center, Nashville, Tennessee

⁹Vanderbilt Eye Institute, Department of Ophthalmology, Vanderbilt University Medical Center, Nashville, Tennessee

¹⁰Roski Eye Institute, Keck School of Medicine, University of Southern California, Los Angeles, California

¹¹Department of Ophthalmology, Mayo Clinic, Rochester, Minnesota

¹²Department of Neurology, Mayo Clinic, Rochester, Minnesota

¹³Byers Eye Institute, Department of Ophthalmology, Stanford University, Palo Alto, California

¹⁴Department of Ophthalmology, University of Washington, Seattle

¹⁵Karalis Johnson Retina Center, Seattle, Washington

¹⁶Byers Eye Institute at Stanford, Stanford University School of Medicine, Palo Alto, California

¹⁷Department of Ophthalmology and Visual Sciences, University of Michigan, Ann Arbor

¹⁸Survey Research Center, Institute for Social Research, University of Michigan, Ann Arbor

¹⁹Department of Ophthalmology and Visual Neurosciences, University of Minnesota, Minneapolis, Minnesota

²⁰Department of Ophthalmology, Mass Eye and Ear and Harvard Medical School, Boston, Massachusetts

²¹Department of Medical Informatics & Clinical Epidemiology, Portland, Oregon

²²Johns Hopkins University School of Medicine, Baltimore, Maryland

²³Tufts University School of Medicine, Boston, Massachusetts

²⁴Department of Clinical Practice, Faculty of Pharmacy, Jazan University, Jazan, Saudi Arabia

²⁵Pharmacy Practice Research Unit, Faculty of Pharmacy, Jazan University, Jazan, Saudi Arabia

²⁶Janssen Research and Development, Titusville, New Jersey

²⁷Department of Biostatistics, UCLA School of Public Health, University of California, Los Angeles, Los Angeles

²⁸VA Informatics and Computing Infrastructure, US Department of Veterans Affairs, Salt Lake City, Utah

²⁹Department of Biostatistics, University of Michigan, Ann Arbor

³⁰Department of Medical Informatics, Erasmus University Medical Center, Rotterdam, the Netherlands

³¹Department of Biomedical Informatics, Columbia University, New York, New York

³²Department of Biostatistics, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland

³³Department of Biomedical Informatics and Data Science, Johns Hopkins School of Medicine, Johns Hopkins University, Baltimore, Maryland

³⁴Department of Internal Medicine, University of Utah School of Medicine, Salt Lake City

³⁵VA Informatics and Computing Infrastructure, US Department of Veterans Affairs, Nashville, Tennessee

³⁶Institute for Informatics, Data Science and Biostatistics, Department of Medicine, Washington University in St Louis, St Louis, Missouri

³⁷Department of Medicine, Johns Hopkins University School of Medicine, Baltimore, Maryland

³⁸Department of Medicine, Washington University in St Louis, St Louis, Missouri

³⁹Department of Internal Medicine, University of California Davis, Sacramento

⁴⁰Department of Medicine, Division of Diabetes and Endocrinology, Vanderbilt University, Nashville, Tennessee

⁴¹Vanderbilt University School of Medicine, Nashville, Tennessee

⁴²Oregon Clinical and Translational Research Institute, Oregon Health & Science University, Portland

⁴³Lee Kong Chian School of Medicine, Nanyang Technological University, Singapore

⁴⁴IQVIA, Real World Solutions, Brighton, United Kingdom

⁴⁵Ruiz Department of Ophthalmology and Visual Science, The University of Texas Health Science Center at Houston, McGovern Medical School, Houston

⁴⁶Department of Medical Informatics and Clinical Epidemiology, Oregon Health & Science University, Portland

⁴⁷Vanderbilt University Medical Center, Nashville, Tennessee

⁴⁸Tan Tock Seng Hospital, Singapore

⁴⁹Stanford School of Medicine and Stanford Health Care, Palo Alto, California

⁵⁰Columbia University Irving Medical Center, New York, New York

⁵¹Johnson & Johnson, Horsham, Pennsylvania

Accepted for Publication: December 8, 2024.

Published Online: February 20, 2025. doi:10.1001/jamaophthalmol.2024.6555

^✉

Corresponding Author: Cindy X. Cai, MD, MS, Wilmer Eye Institute, Johns Hopkins School of Medicine, 1800 Orleans St, Room 711, Baltimore, MD 21287 (ccai6@jhmi.edu).

Author Contributions: Drs Cai and Ryan had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Cai, Hribar, Baxter, Goetz, Swaminathan, Flowers, Brown, Toy, Xu, J. Chen, A. Chen, Wang, C. Lee, Barkmeier, Armbrust, Boland, Boyce, Suchard, Schuemie, Nagy, Mathioudakis, Stocking, L. Lee, Xie, Weiskopf, Ryan.

Acquisition, analysis, or interpretation of data: Cai, Hribar, Baxter, Flowers, Toy, J. Chen, Wang, Leng, Ehrlich, Barkmeier, Boland, Dorr, Alshammari, Swerdel, Suchard, Schuemie, Bu, Sena, Hripcsak, Nishimura, Falconer, DuVall, Matheny, Viernes, O’Brien, Zhang, Martin, Westlund, Mathioudakis, Fan, Wilcox, Lai, Stocking, Takkouche, Humes, McCoy, Adibuzzaman, Areaux, Rojas-Carabali, Brash, D. Lee, Mawn, Agrawal, Morgan-Cooper, Desai, Ryan.

Drafting of the manuscript: Cai, Hribar, Brown, Xu, J. Chen, Wang, Boland, Dorr, Boyce, Alshammari, Swerdel, Suchard, Bu, Sena, Nishimura, Martin, Westlund, Fan, Stocking, L. Lee, Humes, Mawn, Ryan.

Critical review of the manuscript for important intellectual content: Cai, Hribar, Baxter, Goetz, Swaminathan, Flowers, Brown, Toy, Xu, J. Chen, A. Chen, Wang, C. Lee, Leng, Ehrlich, Barkmeier, Armbrust, Boland, Dorr, Alshammari, Swerdel, Suchard, Schuemie, Bu, Hripcsak, Nishimura, Nagy, Falconer, DuVall, Matheny, Viernes, O’Brien, Zhang, Martin, Westlund, Mathioudakis, Wilcox, Lai, Stocking, Takkouche, Xie, Humes, McCoy, Adibuzzaman, Areaux, Rojas-Carabali, Brash, D. Lee, Weiskopf, Mawn, Agrawal, Morgan-Cooper, Desai, Ryan.

Statistical analysis: Toy, Swerdel, Suchard, Schuemie, Bu, Nishimura, Falconer, Viernes, O’Brien, Zhang, Martin, Westlund, Fan, L. Lee, Ryan.

Obtained funding: Hripcsak, DuVall.

Administrative, technical, or material support: Hribar, Baxter, Goetz, Swaminathan, Toy, Wang, Leng, Barkmeier, Dorr, Boyce, Alshammari, Schuemie, Bu, Sena, Hripcsak, Nagy, Falconer, DuVall, Martin, Westlund, Lai, Stocking, Xie, Humes, McCoy, Adibuzzaman, D. Lee, Morgan-Cooper.

Supervision: Cai, Hribar, Goetz, Xu, J. Chen, C. Lee, Leng, Dorr, Suchard, DuVall, Wilcox, Takkouche, Adibuzzaman, Agrawal, Ryan.

Conflict of Interest Disclosures: Dr Cai reported receiving grants from Regeneron Pharmaceuticals outside the submitted work. Dr Baxter reported receiving grants from National Institutes of Health (NIH) and Research to Prevent Blindess, personal fees from Topcon, and nonfinancial support from Optomed outside the submitted work. Dr Swaminathan reported receiving personal fees from AbbVie, Elios Vision, and Lumata Health and grants from Heidelberg Engineering outside the submitted work. Dr Toy reported receiving advisory board fees from Alimera Sciences, Bausch & Lomb, and Regeneron outside the submitted work. Dr C. Lee reported receiving grants from NIH, Centers for Disease Control and Prevention, and the Alzheimer’s Disease Drug Foundation outside the submitted work. Dr Leng reported receiving grants from Astellas and personal fees from Topcon, Astellas, Alcon, Boehringer Ingelheim, Iridex, Nanoscope, Regeneron, Genentech/Roche, Virtual Field, Aptitude Medical, Protagonist Therapeutics, Apellis, and Graybug outside the submitted work. Dr Boland reported receiving personal fees from Carl Zeiss Meditec, Topcon Healthcare, Allergan, and Janssen during the conduct of the study. Dr Swerdel reported being an employee and shareholder of Johnson & Johnson outside the submitted work. Dr Suchard reported receiving grants from NIH, US Department of Veterans Affairs, and the US Food and Drug Administration and personal fees from Johnson & Johnson outside the submitted work. Drs Schuemie and Sena reported being and employee and shareholder of Johnson & Johnson outside the submitted work. Dr DuVall reported receiving grants from Alnylam Pharmaceuticals, Astellas Pharma, AstraZeneca, Biodesix, Celgene Corp, Cerner Enviza, GSK, IQVIA, Janssen Pharmaceuticals, Moderna, Novartis International AG, and Parexel International Corp outside the submitted work. Dr Lai reported being a shareholder of Johnson & Johnson outside the submitted work. Dr Brash reported being employed by IQVIA outside the submitted work. Dr Mawn reported receiving grants from Research to Prevent Blindness and consultant fees from Amgen and Genentech. Dr Ryan reported being an employee and shareholder of Johnson & Johnson. No other disclosures were reported.

Funding/Support: This work was supported in part by grant K23EY033440 from the National Institutes of Health (NIH)/National Eye Institute (NEI) (Dr Cai); grants DP5OD029610 and OT2OD032644 from the NIH (Dr Baxter); grant K23 EY033831 from the NEI (Dr Swaminathan); grant K23 EY032985 from NIH/NEI and grant UL1TR001855 from NIH/National Center for Advancing Translational Science (NCATS) (Dr Toy); grant K12 GCAEI0492E from NIH/NEI (Dr Chen); grant K23 EY03263501 from NIH/NEI (Dr Wang); grants R01AG060942 and OT2OD032644 from NIH (Dr C. Lee); grant P30-EY026877 from NIH (Dr Leng) grant R61AG089063 from X (Dr Ehrlich); this work was supported using resources and facilities of the Department of Veterans Affairs (VA) Informatics and Computing Infrastructure, funded under the research priority to Put VA Data to Work for Veterans (VA ORD 22-D4V); the NIH/National Library of Medicine T15 program (Dr Martin); grant 1K01HL168222 from NIH/National Heart, Lung, and Blood Institute (Dr Stocking); grant UL1TR002369 from NIH/NCATS (Dr Adibuzzaman); grant U24 HD113136-01 from NIH (Dr Nagy); grants R01 HL169954 and R01HL167858 from NIH (Dr Suchard); and grant R01 LM006910 from X (Dr Hripcsak).

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Meeting Presentation: This work was presented at the 2024 OHDSI Community Call; November 19, 2024; virtual.

Data Sharing Statement: See Supplement 3.

^✉

Corresponding author.

PMCID: PMC11843465 PMID: 39976940

This cohort study investigates the association between semaglutide use and nonarteritic anterior ischemic optic neuropathy.

Key Points

Question

Is semaglutide use associated with the nonarteritic anterior ischemic optic neuropathy (NAION)?

Findings

In this large multicenter study of 37.1 million adults with type 2 diabetes across 14 databases in the collaborative Observational Health Data Sciences and Informatics (OHDSI) network, a small increase was identified in the relative incidence of NAION from exposure to semaglutide compared with nonexposure.

Meaning

These findings provide further evidence of an association between semaglutide and NAION but show a smaller risk than that previously reported; additional studies are necessary to identify potential mechanisms and causality.

Abstract

Importance

Semaglutide, a glucagonlike peptide-1 receptor agonist (GLP-1RA), has recently been implicated in cases of nonarteritic anterior ischemic optic neuropathy (NAION), raising safety concerns in the treatment of type 2 diabetes (T2D).

Objective

To investigate the potential association between semaglutide and NAION in the Observational Health Data Sciences and Informatics (OHDSI) network.

Design, Setting, and Participants

This was a retrospective study across 14 databases (6 administrative claims and 8 electronic health records). Included were adults with T2D taking semaglutide, other GLP-1RA (dulaglutide, exenatide), or non–GLP-1RA medications (empagliflozin, sitagliptin, glipizide) from December 1, 2017, to December 31, 2023. The incidence proportion and rate of NAION were calculated. Association between semaglutide and NAION was assessed using 2 approaches: an active-comparator cohort design comparing new users of semaglutide with those taking other GLP-1RAs and non–GLP-1RA drugs, and a self-controlled case-series (SCCS) analysis to compare individuals’ risks during exposure and nonexposure periods for each drug. The cohort design used propensity score–adjusted Cox proportional hazards models to estimate hazard ratios (HRs). The SCCS used conditional Poisson regression models to estimate incidence rate ratios (IRRs). Network-wide HR and IRR estimates were generated using a random-effects meta-analysis model.

Exposures

GLP-1RA and non–GLP-1RAs.

Main Outcomes and Measures

NAION under 2 alternative definitions based on diagnosis codes: one more inclusive and sensitive, the other more restrictive and specific.

Results

The study included 37.1 million individuals with T2D, including 810 390 new semaglutide users. Of the 43 620 new users of semaglutide in the Optum’s deidentified Clinformatics Data Mart Database, 24 473 (56%) were aged 50 to 69 years, and 26 699 (61%) were female. The incidence rate of NAION was 14.5 per 100 000 person-years among semaglutide users. The HR for NAION among new users of semaglutide was not different compared with that of the non–GLP-1RAs using the sensitive NAION definition—empagliflozin (HR, 1.44; 95% CI, 0.78-2.68; P = .12), sitagliptin (HR, 1.30; 95% CI, 0.56-3.01; P = .27), and glipizide (HR, 1.23; 95% CI, 0.66-2.28; P = .25). The risk was higher only compared with patients taking empagliflozin (HR, 2.27; 95% CI, 1.16-4.46; P = .02) using the specific definition. SCCS analysis of semaglutide exposure showed an increased risk of NAION (meta-analysis IRR, 1.32; 95% CI, 1.14-1.54; P < .001).

Conclusions and Relevance

Results of this study suggest a modest increase in the risk of NAION among individuals with T2D associated with semaglutide use, smaller than that previously reported, and warranting further investigation into the clinical implications of this association.

Introduction

Nonarteritic anterior ischemic optic neuropathy (NAION) is the leading cause of acute optic neuropathy in older adults.^1,2,3 A recent single-center study demonstrated a 4.28-fold (95% CI, 1.62-11.29) increased risk of NAION with semaglutide, a glucagonlike peptide-1 receptor agonist (GLP-1RA), relative to non–GLP-1RAs.^4,5 GLP-1RAs, approved by the US Food and Drug Administration (FDA) for type 2 diabetes (T2D) treatment in 2017, offer well-documented benefits in reducing cardiovascular and kidney complications.^{6,7,8,9,10,11} Given these advantages, the American Diabetes Association recommends GLP-1RAs as a preferred therapy for patients with T2D with atherosclerotic cardiovascular disease, chronic kidney disease, or obesity.⁶

There is a need to clarify the potential risk of ophthalmic complications from semaglutide exposure using multicenter data. This study leveraged the Observational Health Data Sciences and Informatics (OHDSI) network to investigate the association between semaglutide exposure and NAION. The OHDSI network is an international open-science collaborative centered around the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM).¹² The OHDSI network and its open-source analytic tools have been previously used to generate reliable evidence in medicine, including ophthalmology.^{7,8,11,12,13,14} We aimed to (1) characterize NAION incidence, (2) compare the risk of NAION associated with semaglutide use against other GLP-1RAs and non–GLP-1RA drugs, and (3) investigate NAION incidence rate during semaglutide exposure compared with nonexposure.

Methods

Study Design

We retrospectively analyzed 14 OHDSI databases (6 administrative claims and 8 electronic health records) using 2 approaches: (1) an active-comparator new-user cohort design to test whether NAION risk is higher with semaglutide compared with other GLP-1RA and non–GLP-1RAs and (2) a self-controlled case-series method to test whether NAION risk was higher with semaglutide exposure compared with nonexposure.^15,16 Database details are in eTable 1 in Supplement 1. All participating sites had local institutional review board approvals where informed consent was waived because this was considered secondary research or exemptions. The study adhered to the tenets of the Declaration of Helsinki. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines.¹⁷

Participants and Exposure

Adults 18 years and older with T2D taking semaglutide (GLP-1RA), dulaglutide (GLP-1RA), exenatide (GLP-1RA), empagliflozin (sodium-glucose cotransporter 2 [SGLT2] inhibitor), sitagliptin (dipeptidyl peptidase 4 [DPP4] inhibitor), or glipizide (sulfonylurea) during the study period (December 1, 2017-December 31, 2023) were included. For the non–GLP-1RA comparator medications, the most commonly used within each drug class was chosen based on the utilization results from the Large-Scale Evidence Generation and Evaluation Across a Network of Databases for Type 2 Diabetes Mellitus (LEGEND-T2DM) study, an OHDSI network study examining second-line antihyperglycemic agents and cardiovascular outcomes.^7,8,11 Participants self-identified races and ethnicities were categorized according to local database guidelines and could include Asian, Black or African American, Hispanic or Latino, not Hispanic or Latino, and White.

Time at Risk

The time-at-risk period began with medication initiation until the end of continuous drug exposure, defined as a gap in exposure of more than 30 days or the end of the continuous observation period.

Outcome

The outcome of interest was NAION, determined by diagnosis codes. A consortium of board-certified ophthalmologists including fellowship-trained neuro-ophthalmologists developed 2 definitions: the sensitive definition required 1 ischemic optic neuropathy diagnosis code, whereas the specific definition required a second confirmatory ischemic optic neuropathy diagnosis code within 90 days. For both definitions, patients with traumatic optic neuropathy or 2 diagnosis codes of giant cell arteritis were excluded. If patients had diagnosis codes within 60 days before the NAION diagnosis related to optic disc disorders, optic neuritis, or optic disc edema, the date of the NAION outcome was shifted to the earlier diagnosis because it was likely related to the NAION event.^18,19,20 Details are in eTables 2 and 3 in Supplement 1 and eTable 4 in Supplement 2.

The sensitivity, specificity, positive predictive value, and negative predictive value of both NAION definitions were evaluated on a subset of databases using PheValuator.²¹ PheValuator is an OHDSI tool that uses machine learning to compute a patient’s probability of having a disease of interest to identify a set of probabilistic gold standards, or patients with high likelihood of having the disease. The probabilistic gold standard dataset was used to determine the performance metrics of the consortium-developed rule-based definitions. Additional details are on the study protocol page.²²

Negative Control Outcomes

In addition to the primary outcome, NAION, we included 97 outcomes believed to be causally unrelated to the exposures under investigation.²² To detect and adjust for systematic errors from residual confounding, both the cohort analysis and self-controlled case-series analysis were performed using these negative control outcomes.^23,24

Statistical Analysis

All analyses were performed in R, version 4.2.3 (R Project for Statistical Computing).^25,26 Analysis codes are available on Github.¹⁶ All P values were 2-sided, and P <.05 was considered statistically significant. Results can also be found on an interactive website.²⁷

Characterization

We summarized the baseline characteristics of patients within each T2DM medication exposure cohort by database.²⁸ These characteristics included demographics (age, sex, race, and ethnicity), Diabetes Comorbidity Severity Index score, Charlson Comorbidity Index-Romano adaptation, prior medical diagnoses (essential hypertension, hyperlipidemia, obstructive sleep apnea, chronic kidney disease, and anemia), and prior medication use (interferon, amiodarone, and phosphodiesterase inhibitor).^29,30 NAION incidence proportion (number of patients with NAION divided by persons at risk) and incidence rate (number of patients with NAION during time-at-risk divided by person-days) were calculated.^28,31

Active-Comparator New-User Cohort Analysis

An active-comparator new-user cohort design was used to estimate propensity score–adjusted hazard ratios (HRs) for NAION comparing new semaglutide users with users of comparator medications: GLP-1RAs (dulaglutide and exenatide) and non–GLP-1RA drugs (empagliflozin, sitagliptin, and glipizide).³² New users were defined as previously described.^7,8,11 In brief, adults with T2D taking metformin monotherapy were included if they had at least 1 year of prior observation, initiated treatment with one of the medications of interest, had no prior exposure to a comparator diabetes medication, and had at most 30 days of prior insulin use.

Propensity scores were estimated using a large-scale propensity score approach, applying regularized regression over all baseline characteristics (eg, demographic characteristics, preexisting conditions, medications, and procedures).^33,34 Patients in each target and comparator exposure comparison (eg, semaglutide vs dulaglutide) were matched 1:1 using propensity scores. Cox proportional hazards models estimated the HR of NAION from cohort entry to the outcome while taking treatment with each target and comparator T2D medication.

Sensitivity Analyses

We used 2 sensitivity analyses to test robustness. To assess sensitivity to the cohort definition, we created a second definition to include new users of each T2D medication regardless of prior exposure to a comparator drug. Details can be found in eTables 2 and 3 in Supplement 1 and eTable 4 in Supplement 2.²² We also stratified our propensity score–matched Cox proportional hazards models by calendar time (December 2017-January 2020, February 2020-June 2021, and July 2021-December 2023). This addressed potential biases due to decreased health care utilization from the COVID-19 pandemic, and the 60% increased prescription of semaglutide from 2021 to 2023 after FDA approval for obesity.^22,35

Self-Controlled Case-Series Analysis

A self-controlled case series, where patients served as their own controls, was used to estimate the NAION incidence rate ratio (IRR) during each T2D medication exposure compared with nonexposure control time.^36,37,38 Real-world data evaluation of study designs have found that this approach can accurately identify causal associations between drug exposures and health outcomes.^36,38,39 We restricted the observation period to when patients had T2D and excluded the patients’ first 365 days in the database from the analysis to improve the identification of incident NAION. We used conditional Poisson regression models to compare the NAION IRRs.⁴⁰ We defined a separate preexposure time window, during the control time, as the period of 30 days before treatment initiation and adjusted for it.^40,41 The models also adjusted for potential effects of seasonality by including spline functions of calendar months.

Study Diagnostics and Meta-Analysis

For both cohort and self-controlled case-series analyses, we used a comprehensive set of study diagnostics, eg, preexposure patient characteristic balance, empirical clinical equipoise, minimum power requirement, and negative control experiments, to evaluate systematic error, residual bias, reliability, and the generalizability of our comparisons.^{23,24,36,42,43} Only databases that passed prespecified study diagnostics reported HR or IRR estimates.²² A bayesian random-effects meta-analysis combined each database’s estimates into a network-wide estimate.^44,45

Results

In total, 14 OHDSI network databases were included in the analysis; all 14 were included in the NAION incidence analysis, but only 8 in the active-comparator new-user cohort analysis and 10 in the self-controlled case-series analysis passed study diagnostics (eTable 1 in Supplement 1). Results from the diagnostics are shown in eTables 5 to 8 in Supplement 1.

PheValuator confirmed that the sensitive NAION definition has higher sensitivities (mean 73% vs 33%), whereas the specific NAION definition has higher positive predictive values (88% vs 71%) (eTable 9 in Supplement 1).

NAION Incidence

Across all databases, there were 37.1 million patients with T2D, among whom 810 390 were new semaglutide users, 326 282 dulaglutide, 25 936 exenatide, 715 802 empagliflozin, 493 563 sitagliptin, and 832 295 glipizide. Baseline patient characteristics in the Optum’s deidentified Clinformatics Data Mart Database are presented in the Table, as a representative example. There were a total of 166 932 patients in all T2D drug exposure cohorts, with 43 620 new users of semaglutide, of whom 24 473 (56%) were aged 50 to 69 years, 26 699 (61%) were female, and 16 921 (39%) were male. Patients self-identified with the following race ethnicities: 1179 Asian (3%), 5748 Black or African American (13%), 4920 Hispanic (11%), 35 562 not Hispanic or Latino (82%), and 28 635 White (66%). The baseline patient characteristics of all other databases are in eTable 10 in Supplement 1.

Table. Baseline Characteristics of Patients in Each Type 2 Diabetes Drug Exposure Cohort (Semaglutide, Dulaglutide, Exenatide, Empagliflozin, Sitagliptin, Glipizide) During the Study Period in the Optum’s Deidentified Clinformatics Data Mart Database (Clinformatics).

Variable	Patients, No. (%)
	GLP-1 RA			Empagliflozin (SGLT2 inhibitor)	Sitagliptin (DPP4 inhibitor)	Glipizide (sulfonylurea)
	Semaglutide	Dulaglutide	Exenatide	Empagliflozin (SGLT2 inhibitor)	Sitagliptin (DPP4 inhibitor)	Glipizide (sulfonylurea)
No.	43 620	14 923	1414	43 302	19 581	44 092
Age, y^a
≤29	453 (1)	168 (1)	23 (1)	165 (0)	29 (0)	174 (0)
30-49	8110 (19)	3095 (21)	315 (22)	5122 (12)	875 (4)	4478 (10)
50-69	24 473 (56)	8238 (55)	832 (59)	21 614 (50)	7292 (37)	20 330 (46)
≥70	10 586 (24)	3426 (23)	248 (18)	16 401 (38)	11 389 (58)	19 111 (43)
Race
Asian	1179 (3)	332 (2)	37 (3)	2333 (5)	961 (5)	2008 (5)
Black or African American	5748 (13)	1939 (13)	166 (12)	5228 (12)	3020 (15)	5458 (12)
White	28 635 (66)	9976 (67)	900 (64)	27 337 (63)	11 283 (58)	26 464 (60)
Ethnicity
Hispanic or Latino	4920 (11)	1703 (11)	228 (16)	5329 (12)	2892 (15)	6836 (16)
Not Hispanic or Latino	35 562 (82)	12 247 (82)	1103 (78)	34 898 (81)	15 264 (78)	33 930 (77)
Sex
Female	26 699 (61)	8343 (56)	797 (56)	17 872 (41)	10 683 (55)	20 587 (47)
Male	16 921 (39)	6580 (44)	617 (44)	25 430 (59)	8898 (45)	23 505 (53)
DCSI, mean (SD)	2.57 (1.97)	2.64 (1.97)	2.72 (2.08)	3.25 (2.25)	3.57 (2.39)	3.26 (2.31)
CCI, mean (SD)	4.04 (2.37)	4.05 (2.33)	3.92 (2.23)	4.69 (2.67)	5.03 (2.71)	4.64 (2.68)
Essential hypertension	33 537 (77)	11 420 (77)	1087 (77)	34 789 (80)	16 441 (84)	35 070 (80)
Hyperlipidemia	34 056 (78)	11 393 (76)	1079 (76)	35 942 (83)	16 342 (83)	34 918 (79)
Obstructive sleep apnea	12 079 (28)	3729 (25)	348 (25)	8654 (20)	2905 (15)	6142 (14)
Chronic kidney disease	4776 (11)	1670 (11)	134 (9)	7358 (17)	4142 (21)	8067 (18)
Anemia	6373 (15)	1859 (12)	172 (12)	7048 (16)	3888 (20)	7178 (16)
Interferon use	6 (0)	5 (0)	0	7 (0)	<5 (0)	9 (0)
Amiodarone use	218 (0)	89 (1)	10 (1)	840 (2)	248 (1)	473 (1)
Phosphodiesterase inhibitor use	0	0	0	6 (0)	<5 (0)	<5 (0)

Open in a new tab

Abbreviations: CCI, Charlson Comorbidity Index-Romano adaptation; DCSI, Diabetes Complications Severity Index; DPP4, dipeptidyl peptidase-4; GLP-1 RA, glucagonlike peptide 1 receptor agonist; SGLT2, sodium-glucose cotransporter 2.

^{^a}

In each database, age groups span 5 years, and counts fewer than 5 are reported as <5. When combining age groups for reporting, all such categories were treated as having 5 patients.

Among new semaglutide users, the NAION incidence proportions were 7.1 and 4.2 per 100 000 persons, and the incidence rates were 14.5 and 8.7 per 100 000 person-years, for the sensitive and specific definitions, respectively (Figure 1 and eTable 11 in Supplement 1).

Active-Comparator New-User Cohort Analysis

The risk for NAION using the sensitive definition was not different among new semaglutide users compared with dulaglutide (meta-analysis HR, 0.93; 95% CI, 0.46-1.91; P = .57) or with the non–GLP-1RA medications, empagliflozin (HR, 1.44; 95% CI, 0.78-2.68; P = .12), sitagliptin (HR, 1.30; 95% CI, 0.56-3.01; P = .27), and glipizide (HR, 1.23; 95% CI, 0.66-2.28; P = .25) (Figure 2 and eFigures 1 and 2 in Supplement 1). Similarly, using the specific definition, the risk was not different among new users of semaglutide compared with dulaglutide (HR, 1.28; 95% CI, 0.57-2.88; P = .55), sitagliptin (HR, 1.64; 95% CI, 0.69-3.90; P = .27), and glipizide (HR, 1.50; 95% CI, 0.70-3.21; P = .29). There was, however, increased NAION risk among new semaglutide users compared with empagliflozin (HR, 2.27; 95% CI, 1.59-4.46; P = .02) using the specific definition (Figure 2).

Figure 2. — Hazard ratio (HR) and 95% CI estimates for the risk of nonarteritic anterior ischemic optic neuropathy (NAION) while receiving treatment with a second-line medication for type 2 diabetes comparing between semaglutide and other glucagonlike peptide-1 receptor agonists (GLP-1RAs), ie, dulaglutide and exenatide, and non–GLP-1RA medications, ie, empagliflozin, sitagliptin, and glipizide. Only results from databases and comparisons that passed study diagnostics are provided, as well as the meta-analytic estimates.^a The total number of patients in each comparison is shown. A, Results using the sensitive NAION definition. B, Results using the specific NAION definition.^bCCAE indicates Merative MarketScan Commercial Claims and Encounters Database; Clinformatics, Optum’s deidentified Clinformatics Data Mart Database; IQVIA, IQVIA Open Claims; PharMetrics, PharMetrics Plus.

^aDatabases that have 0 cases of NAION are unable to produce HR estimates but still contribute to the meta-analytic estimate.

^bThe sensitive NAION definition only required 1 diagnosis of ischemic optic neuropathy, and the specific NAION definition required 2 diagnoses. All other criteria were the same between the 2 definitions.

Results were similar in the 2 sensitivity analyses (eFigures 3 and 4 in Supplement 1). The risk of NAION was higher among semaglutide users regardless of prior medication exposure compared with empagliflozin (both definitions) and with sitagliptin (specific definition). NAION risk was also higher among semaglutide users from 2021 to 2023 compared with empagliflozin using both NAION definitions.

Self-Controlled Case-Series Analysis

Up to 39 104 patients with NAION for the sensitive and 22 005 for the specific definitions contributed to this analysis. There was an increased NAION incidence rate associated with semaglutide exposure using both the sensitive (meta-analysis IRR, 1.32; 95% CI, 1.14-1.54; P < .001) (Figure 3) and the specific definitions (IRR, 1.50; 95% CI, 1.26-1.79; P < .001) (Figure 4). For exenatide, there was an increased incidence rate using the specific definition (IRR, 1.62; 95% CI, 1.02-2.58; P = .04) (Figure 4) but not the sensitive definition (IRR, 1.26; 95% CI, 0.78-2.03; P = .35) (Figure 3). There was no increased NAION incidence rate associated with dulaglutide or any non–GLP-1RA (Figure 3 and Figure 4).

Figure 3. — Incidence rate ratio (IRR) and 95% CI estimate for the risk of NAION (using the sensitive definition)^a while receiving treatment with one of the medications for type 2 diabetes compared with control time, not taking treatment with the medication of interest. Results are shown for semaglutide, other glucagonlike peptide-1 receptor agonists (GLP-1 RAs), ie, dulaglutide and exenatide, and non–GLP-1 RA medications, ie, empagliflozin, sitagliptin, and glipizide. Results from databases that passed study diagnostics are provided, as well as the meta-analytic estimates.^b The total number of patients included is shown. CCAE indicates Merative MarketScan Commercial Claims and Encounters Database; Clinformatics, Optum’s deidentified Clinformatics Data Mart; IQVIA, IQVIA Open Claims; JHME, Johns Hopkins Medical Enterprise; MDCD, Merative MarketScan Multi-State Medicaid Database; MDCR, Merative MarketScan Medicare Supplemental and Coordination of Benefits Database; PharMetrics, PharMetrics Plus; VA, Department of Veterans Affairs; WashU, Washington University in St Louis.

^aThe sensitive NAION definition only required 1 diagnosis of ischemic optic neuropathy. All other criteria were the same between the 2 definitions.

^bDue to data privacy issues, databases that have counts less than 5 are unable to produce IRR estimates but still contribute to the meta-analytic estimate.

Figure 4. — Incidence rate ratio (IRR) and 95% CI estimate for the risk of NAION (using the specific definition)^a while receiving treatment with one of the medications for type 2 diabetes compared with control time, not taking treatment with the medication of interest. Results are shown for semaglutide, other glucagonlike peptide-1 receptor agonists (GLP-1 RAs), ie, dulaglutide and exenatide, and non–GLP-1 RA medications, ie, empagliflozin, sitagliptin, and glipizide. Results from databases that passed study diagnostics are provided, as well as the meta-analytic estimates.^b The total number of patients included is shown. CCAE indicates Merative MarketScan Commercial Claims and Encounters Database; Clinformatics, Optum’s deidentified Clinformatics Data Mart; IQVIA, IQVIA Open Claims; JHME, Johns Hopkins Medical Enterprise; MDCD, Merative MarketScan Multi-State Medicaid Database; MDCR, Merative MarketScan Medicare Supplemental and Coordination of Benefits Database; PharMetrics, PharMetrics Plus; VA, Department of Veterans Affairs; WashU, Washington University in St Louis.

^aThe specific NAION definition required 2 diagnoses. All other criteria were the same between the 2 definitions.

^bDue to data privacy issues, databases that have counts less than 5 are unable to produce IRR estimates but still contribute to the meta-analytic estimate.

Discussion

In this retrospective study across 14 OHDSI network databases evaluating 37.1 million patients with T2D, the NAION incidence rate among new semaglutide users was 14.5 and 8.7 per 100 000 person-years (sensitive and specific NAION definitions, respectively). In the cohort analysis, we observed increased NAION risk comparing propensity score–matched new semaglutide users to users of non–GLP-1RA drugs empagliflozin and sitagliptin, depending on the NAION definition. In the self-control case-series analysis, we observed an increased NAION risk with semaglutide exposure relative to nonexposure; this finding was consistent across both NAION definitions. We also observed an increased NAION risk among patients exposed to exenatide, another GLP-1RA but only using the specific NAION definition. No increased risk was observed for the other GLP-1RA (dulaglutide), or non–GLP-1RA medications (empagliflozin, sitagliptin, glipizide).

Historically, the annual incidence of NAION has been reported as 2.3 to 11.4 per 100 000 persons among all patients depending on age group.^{46,47,48,49,50,51} However, some estimates have been as high as 82 per 100 000 persons, among individuals older than 68 years.⁵² The NAION incidence among patients with T2D in our study is within this range. Among new users of second-line T2D medications, our estimates for the NAION incidence proportion are much lower than estimates of 8.9% for semaglutide and 1.8% for non–GLP-1RAs from Hathaway et al.⁴ This is likely due to study population differences; Hathaway et al⁴ focused on neuro-ophthalmology patients at a major NAION referral center, whereas our population was from national databases across the US.

We created 2 definitions to address the lack of structured diagnostic codes specific to NAION. Manual records review suggests that 40% of cases coded as ischemic optic neuropathy are not actually NAION.⁴ Because manual review is not feasible for large database studies, we used 2 definitions of NAION developed through expert consensus to test the robustness of our findings against case misspecification. Because a definitive diagnosis often requires multiple visits, we improved specificity by requiring a second ischemic optic neuropathy diagnosis code in one of the definitions.^19,20,53,54 Although the positive predictive value of the diagnosis codes can be enhanced with inclusion of the practitioner specialty, we were unable to include practitioner specialty due to limitations of the databases.⁵³ The difference in the specificity of the 2 NAION definitions explains the variability in our estimated incidences.

Our study findings are largely consistent across the 2 analyses and NAION definitions. In the active-comparator new-user cohort analysis, the meta-analysis estimates of HRs ranged from 1.23 to 2.27 for new semaglutide users compared with non–GLP-1RA medications. The findings, however, were only in the comparisons with empagliflozin in the primary analysis and sitagliptin in a sensitivity analysis, both using the specific NAION definition. Case misspecification could be playing a role in the lack of statistical significance using the sensitive NAION definition. Although we did not adjust for multiple comparisons in the cohort analysis, findings were consistent in the self-controlled case series. We found an increase in NAION risks from semaglutide exposure compared with nonexposure, with an IRR of 1.32 and 1.50 using both specific and sensitive definitions. There is more statistical power in the self-controlled case series analysis since more databases passed study diagnostics and thus more patients were included.

Two prior studies^4,5 have investigated NAION and semaglutide and have found contradictory findings. We do not know if the differences were due to variations in study design, patient population, statistical power, or some combination. The ability to assess how results vary across study designs and databases is an important feature and advantage of OHDSI network studies.^{7,8,11,12,13,55} Our self-controlled case-series findings for semaglutide also varied among databases—one-half the contributing databases did not demonstrate an increased IRR of NAION with semaglutide exposure. If our study had been performed on any of those databases alone, we would not have detected an increased IRR of NAION.

It is unclear if the increased risk of NAION is specific to semaglutide, or also true of other GLP-1RAs. In our study, we did not see an increased risk of NAION associated with dulaglutide. However, we did find a small increased risk of NAION with exenatide use when using the specific NAION definition. There are no known mechanistic links between GLP-1RA and NAION.⁴ On the contrary, GLP-1RAs have been associated with neuroprotective properties and reduction of ischemic risk.^56,57,58 It is possible that GLP-1RAs affect vascular dynamics via the autonomic nervous system and lower systemic blood pressure, which could affect optic nerve head perfusion and increase NAION risk.^4,9,59,60

Strengths and Limitations

This study has some strengths. This was, to our knowledge, the largest study to investigate the association of NAION with semaglutide and other GLP-1RA and non–GLP-1RA T2D medications. Our sample size allowed us to identify many more NAION cases than prior studies. The diversity of databases allowed us to estimate incidence rates nationally. The rigorous study diagnostics addressed many biases and is another strength.

There were also some limitations with our study. Given lack of access to source records, we were unable to definitively distinguish incident cases of NAION from prevalent cases. However, the prior observation period in the cohort design and exclusion of the first year of observation in the self-controlled case series analysis increased the likelihood that the cases represented incident NAION. Some databases did not have patients older than 65 years; therefore, we could be missing cases of NAION as it often affects an older population. We were also unable to include cup-disc ratio data. Crowded optic nerves (ie, a small cup-disc ratio) is the strongest risk factor for NAION, outweighing systemic risk factors like diabetes.⁶¹ Incorporating ophthalmic examination data into the OMOP CDM is an active pursuit by the OHDSI Eye Care and Vision Research Workgroup.^62,63 We were also unable to evaluate any heterogeneity of effects, such as whether semaglutide presents a greater risk for NAION among patients with anatomic predisposition or more severe diabetes (eg, by duration or hemoglobin A_1c level) or whether risk differs for patients who take semaglutide for obesity. In the OMOP CDM, drugs are identified by their active ingredients; the semaglutide group, therefore, included the subcutaneous injection (Ozempic [Novo Nordisk]) and by-mouth formulation (Rybelsus [Novo Nordisk]). The subcutaneous form of semaglutide has greater bioavailability, shows greater efficacy in glucose lowering, and is more commonly prescribed.^35,64 Even with the inclusion of a less effective semaglutide formulation we were able to identify an increased risk of NAION. We were also unable to include drug dose in the analysis. Time-varying confounding in the self-controlled case series is a possible concern. Diabetes worsens over time with aging, increasing likelihood of both being prescribed T2D medications and of developing NAION. However, this should affect the analysis of all T2D medications; the higher estimates of NAION risk under semaglutide relative to other medications suggests our finding not to be an artifact of such time-varying confounding. Patients with T2D who develop coronary artery disease are more likely to be prescribed semaglutide, contributing to potential indication bias because coronary artery disease is also an NAION risk factor.⁶⁵ To alleviate this concern, we excluded databases where there was a higher NAION incidence during the 30-day preexposure period, which would suggest such indication bias. Databases showing substantial systematic error on the negative control outcomes, expressed as an absolute expected systematic error greater than 0.25, were also excluded.

Conclusions

In conclusion, in this retrospective study, we identified a small increased risk of NAION among patients with T2D who were exposed to semaglutide, smaller than that previously reported.⁴ Additional studies that incorporate ophthalmic risk factors or examine dose-dependent effects, eg, are needed to investigate a potential causal relationship between semaglutide and NAION. In the absence of a known mechanism for this association, we urge clinicians to weigh the concern for an increased risk of a rare but potentially blinding eye condition with the many therapeutic benefits of semaglutide.

Supplement 1.

eTable 1. Description of the Databases (Administrative Claims and Electronic Health Records) Included in the Study

eTable 2. Details of the Indication Cohort (Patients With T2DM), Exposure Cohorts (Semaglutide, Dulaglutide, Exenatide, Empagliflozin, Sitagliptin, Glipizide), and Outcome Cohorts (Sensitive and Specific Definitions of NAION)

eTable 3. Codes From Standardized Vocabularies Used to Define Clinical Concepts

eTable 5. Results From the Study Diagnostics for the New-User Active-Comparator Cohort Design Analysis for the “Sensitive” NAION Definition

eTable 6. Results From the Study Diagnostics for the New-User Active-Comparator Cohort Design Analysis for the “Specific” NAION Definition

eTable 7. Results from the Study Diagnostics for the Self-Controlled Case-Series Analysis for the “Sensitive” NAION Definition

eTable 8. Results From the Study Diagnostics for the Self-Controlled Case-Series Analysis for the “Specific” NAION Definition

eTable 9. Performance Characteristics of Definition Algorithms for the “Sensitive” and “Specific” Definitions of NAION

eTable 10. Baseline Characteristics of Patients in Each Exposure Cohort Semaglutide, Dulaglutide, Exenatide, Empagliflozin, Sitagliptin, Glipizide During the Study Period in All Databases

eTable 11. Incidence Proportion and Incidence Rate of NAION Among Adults With T2DM and in Each T2DM Drug Exposure Cohort (Semaglutide, Dulaglutide, Exenatide, Empagliflozin, Sitagliptin, Glipizide) Across All Databases

eFigure 1. Kaplan-Meier Plot Showing Survival as a Function of Time Comparing New Users of Semaglutide and Dulaglutide Using the “Sensitive” NAION Definition

eFigure 2. Kaplan-Meier Plot Showing Survival as a Function of Time Comparing New Users of Semaglutide and Empagliflozin Using the “Sensitive” NAION Definition

eFigure 3. Forest Plot for Active-Comparator Cohort Design, Results From the First Sensitivity Analysis (Not Requiring Second-Line Treatment)

eFigure 4. Forest Plot for Active-Comparator Cohort Design, Results From the Second Sensitivity Analysis (Restriction by Calendar Time)

jamaophthalmol-e246555-s001.pdf^{(1.8MB, pdf)}

Supplement 2.

eTable 4. Nonstandard OMOP to Standard OMOP Concept Mapping

jamaophthalmol-e246555-s002.xlsx^{(68.8KB, xlsx)}

Supplement 3.

Data Sharing Statement.

jamaophthalmol-e246555-s003.pdf^{(13.1KB, pdf)}

Journal Club Slides

jamaophthalmol-e246555-slides.pptx^{(655.8KB, pptx)}

References

1.Hayreh SS. Ischemic optic neuropathy. Prog Retin Eye Res. 2009;28(1):34-62. doi: 10.1016/j.preteyeres.2008.11.002 [DOI] [PubMed] [Google Scholar]
2.Hayreh SS, Zimmerman MB. Nonarteritic anterior ischemic optic neuropathy: natural history of visual outcome. Ophthalmology. 2008;115(2):298-305.e2. doi: 10.1016/j.ophtha.2007.05.027 [DOI] [PMC free article] [PubMed] [Google Scholar]
3.The Ischemic Optic Neuropathy Decompression Trial Research Group . Optic nerve decompression surgery for nonarteritic anterior ischemic optic neuropathy (NAION) is not effective and may be harmful. JAMA. 1995;273(8):625-632. doi: 10.1001/jama.1995.03520320035038 [DOI] [PubMed] [Google Scholar]
4.Hathaway JT, Shah MP, Hathaway DB, et al. Risk of nonarteritic anterior ischemic optic neuropathy in patients prescribed semaglutide. JAMA Ophthalmol. 2024;142(8):732-739. doi: 10.1001/jamaophthalmol.2024.2296 [DOI] [PMC free article] [PubMed] [Google Scholar]
5.Klonoff DC, Hui G, Gombar S. Real-world evidence assessment of the risk of nonarteritic anterior ischemic optic neuropathy in patients prescribed semaglutide. J Diabetes Sci Technol. 2024;18(6):1517-1518. doi: 10.1177/19322968241268050 [DOI] [PMC free article] [PubMed] [Google Scholar]
6.American Diabetes Association Professional Practice Committee . 9. Pharmacologic approaches to glycemic treatment: standards of care in diabetes—2024. Diabetes Care. 2024;47(suppl 1):S158-S178. doi: 10.2337/dc24-S009 [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Khera R, Schuemie MJ, Lu Y, et al. Large-scale evidence generation and evaluation across a network of databases for type 2 diabetes mellitus (LEGEND-T2DM): a protocol for a series of multinational, real-world comparative cardiovascular effectiveness and safety studies. BMJ Open. 2022;12(6):e057977. doi: 10.1136/bmjopen-2021-057977 [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Khera R, Aminorroaya A, Dhingra LS, et al. Comparative effectiveness of second-line antihyperglycemic agents for cardiovascular outcomes: a large-scale, multinational, federated analysis of the LEGEND-T2DM study. medRxiv. Preprint posted online February 8, 2024. doi: 10.1101/2024.02.05.24302354 [DOI] [PMC free article] [PubMed]
9.Marso SP, Bain SC, Consoli A, et al. ; SUSTAIN-6 Investigators . Semaglutide and cardiovascular outcomes in patients with type 2 diabetes. N Engl J Med. 2016;375(19):1834-1844. doi: 10.1056/NEJMoa1607141 [DOI] [PubMed] [Google Scholar]
10.Perkovic V, Tuttle KR, Rossing P, et al. ; FLOW Trial Committees and Investigators . Effects of semaglutide on chronic kidney disease in patients with type 2 diabetes. N Engl J Med. 2024;391(2):109-121. doi: 10.1056/NEJMoa2403347 [DOI] [PubMed] [Google Scholar]
11.Khera R, Aminorroaya A, Dhingra LS, et al. Comparative effectiveness of second-line antihyperglycemic agents for cardiovascular outcomes: a multinational, federated analysis of LEGEND-T2DM. J Am Coll Cardiol. 2024;84(10):904-917. doi: 10.1016/j.jacc.2024.05.069 [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Hripcsak G, Duke JD, Shah NH, et al. Observational Health Data Sciences and Informatics (OHDSI): opportunities for observational researchers. Stud Health Technol Inform. 2015;216:574-578. [PMC free article] [PubMed] [Google Scholar]
13.Cai CX, Nishimura A, Bowring MG, et al. Similar risk of kidney failure among patients with blinding diseases who receive ranibizumab, aflibercept, and bevacizumab: an Observational Health Data Sciences and Informatics network study. Ophthalmol Retina. 2024;8(8):733-743. doi: 10.1016/j.oret.2024.03.014 [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Suchard MA, Schuemie MJ, Krumholz HM, et al. Comprehensive comparative effectiveness and safety of first-line antihypertensive drug classes: a systematic, multinational, large-scale analysis. Lancet. 2019;394(10211):1816-1826. doi: 10.1016/S0140-6736(19)32317-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Overhage JM, Ryan PB, Reich CG, Hartzema AG, Stang PE. Validation of a common data model for active safety surveillance research. J Am Med Inform Assoc. 2012;19(1):54-60. doi: 10.1136/amiajnl-2011-000376 [DOI] [PMC free article] [PubMed] [Google Scholar]
16.GitHub . OHDSI studies/semaglutide naion. Accessed August 17, 2024. https://github.com/ohdsi-studies/SemaglutideNaion
17.von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP; STROBE Initiative . Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. BMJ. 2007;335(7624):806-808. doi: 10.1136/bmj.39335.541782.AD [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Stunkel L, Kung NH, Wilson B, McClelland CM, Van Stavern GP. Incidence and causes of overdiagnosis of optic neuritis. JAMA Ophthalmol. 2018;136(1):76-81. doi: 10.1001/jamaophthalmol.2017.5470 [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Stunkel L, Newman NJ, Biousse V. Diagnostic error and neuro-ophthalmology. Curr Opin Neurol. 2019;32(1):62-67. doi: 10.1097/WCO.0000000000000635 [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Stunkel L, Sharma RA, Mackay DD, et al. Patient harm due to diagnostic error of neuro-ophthalmologic conditions. Ophthalmology. 2021;128(9):1356-1362. doi: 10.1016/j.ophtha.2021.03.008 [DOI] [PubMed] [Google Scholar]
21.Swerdel JN, Schuemie M, Murray G, Ryan PB. PheValuator 2.0: methodological improvements for the PheValuator approach to semi-automated phenotype algorithm evaluation. J Biomed Inform. 2022;135:104177. doi: 10.1016/j.jbi.2022.104177 [DOI] [PubMed] [Google Scholar]
22.OHDSI . Research protocol. Accessed August 17, 2024. https://ohdsi-studies.github.io/SemaglutideNaion/protocol.html
23.Schuemie MJ, Ryan PB, DuMouchel W, Suchard MA, Madigan D. Interpreting observational studies: why empirical calibration is needed to correct P values. Stat Med. 2014;33(2):209-218. doi: 10.1002/sim.5925 [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Schuemie MJ, Hripcsak G, Ryan PB, Madigan D, Suchard MA. Empirical confidence interval calibration for population-level effect estimation studies in observational healthcare data. Proc Natl Acad Sci U S A. 2018;115(11):2571-2577. doi: 10.1073/pnas.1708282114 [DOI] [PMC free article] [PubMed] [Google Scholar]
25.GitHub . OHDSI/strategus. Accessed August 17, 2024. https://github.com/OHDSI/Strategus
26.Schuemie M, Reps J, Black A, et al. Health-Analytics Data to Evidence Suite (HADES): open-source software for observational research. Stud Health Technol Inform. 2024;310:966-970. doi: 10.3233/SHTI231108 [DOI] [PMC free article] [PubMed] [Google Scholar]
27.OHDSI Evidence Sharing . OHDSI analysis viewer. Accessed September 5, 2024. https://results.ohdsi.org/app/20_Semanaion_Char
28.GitHub . OHDSI/characterization. Accessed August 17, 2024. https://github.com/OHDSI/Characterization
29.Glasheen WP, Renda A, Dong Y. Diabetes Complications Severity Index (DCSI)—update and ICD-10 translation. J Diabetes Complications. 2017;31(6):1007-1013. doi: 10.1016/j.jdiacomp.2017.02.018 [DOI] [PubMed] [Google Scholar]
30.Romano PS, Roos LL, Jollis JG. Adapting a clinical comorbidity index for use with ICD-9-CM administrative data: differing perspectives. J Clin Epidemiol. 1993;46(10):1075-1079. doi: 10.1016/0895-4356(93)90103-8 [DOI] [PubMed] [Google Scholar]
31.GitHub . OHDSI/cohort incidence. Accessed August 17, 2024. https://github.com/OHDSI/CohortIncidence
32.GitHub . OHDSI/cohort method. Accessed August 17, 2024. https://github.com/OHDSI/CohortMethod
33.Tian Y, Schuemie MJ, Suchard MA. Evaluating large-scale propensity score performance through real-world and synthetic data experiments. Int J Epidemiol. 2018;47(6):2005-2014. doi: 10.1093/ije/dyy120 [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Zhang L, Wang Y, Schuemie MJ, Blei DM, Hripcsak G. Adjusting for indirectly measured confounding using large-scale propensity score. J Biomed Inform. 2022;134:104204. doi: 10.1016/j.jbi.2022.104204 [DOI] [PMC free article] [PubMed] [Google Scholar]
35.Novo Nordisk . Q1 2023 presentation. Accessed August 18, 2024. https://investor.novonordisk.com/q1-2023-presentation/?page=15
36.Schuemie MJ, Cepeda MS, Suchard MA, et al. How confident are we about observational findings in health care: a benchmark study. Harv Data Sci Rev. 2020;2(1). doi: 10.1162/99608f92.147cc28e [DOI] [PMC free article] [PubMed] [Google Scholar]
37.GitHub . OHDSI/self-controlled case series. Accessed August 17, 2024. https://github.com/OHDSI/SelfControlledCaseSeries
38.Ryan PB, Schuemie MJ. Evaluating performance of risk identification methods through a large-scale simulation of observational data. Drug Saf. 2013;36(suppl 1):S171-S180. doi: 10.1007/s40264-013-0110-2 [DOI] [PubMed] [Google Scholar]
39.Madigan D, Stang PE, Berlin JA, et al. A systematic statistical approach to evaluating evidence from observational studies [abstract]. Annu Rev Stat Appl. 2014;1(1):11-39. doi: 10.1146/annurev-statistics-022513-115645 [DOI] [Google Scholar]
40.Whitaker HJ, Farrington CP, Spiessens B, Musonda P. Tutorial in biostatistics: the self-controlled case series method. Stat Med. 2006;25(10):1768-1797. doi: 10.1002/sim.2302 [DOI] [PubMed] [Google Scholar]
41.Petersen I, Douglas I, Whitaker H. Self-controlled case series methods: an alternative to standard epidemiological study designs. BMJ. 2016;354:i4515. doi: 10.1136/bmj.i4515 [DOI] [PubMed] [Google Scholar]
42.Schuemie MJ, Ryan PB, Hripcsak G, Madigan D, Suchard MA. Improving reproducibility by using high-throughput observational studies with empirical calibration. Philos Trans A Math Phys Eng Sci. 2018;376(2128):20170356. doi: 10.1098/rsta.2017.0356 [DOI] [PMC free article] [PubMed] [Google Scholar]
43.Schuemie MJ, Hripcsak G, Ryan PB, Madigan D, Suchard MA. Robust empirical calibration of P values using observational data. Stat Med. 2016;35(22):3883-3888. doi: 10.1002/sim.6977 [DOI] [PMC free article] [PubMed] [Google Scholar]
44.Schuemie MJ, Chen Y, Madigan D, Suchard MA. Combining cox regressions across a heterogeneous distributed research network facing small and zero counts. Stat Methods Med Res. 2022;31(3):438-450. doi: 10.1177/09622802211060518 [DOI] [PubMed] [Google Scholar]
45.GitHub . OHDSI/evidence synthesis. Accessed August 17, 2024. https://github.com/OHDSI/EvidenceSynthesis
46.Johnson LN, Arnold AC. Incidence of nonarteritic and arteritic anterior ischemic optic neuropathy: population-based study in the state of Missouri and Los Angeles County, California. J Neuroophthalmol. 1994;14(1):38-44. doi: 10.1097/00041327-199403000-00011 [DOI] [PubMed] [Google Scholar]
47.Hattenhauer MG, Leavitt JA, Hodge DO, Grill R, Gray DT. Incidence of nonarteritic anterior ischemic optic neuropathy. Am J Ophthalmol. 1997;123(1):103-107. doi: 10.1016/S0002-9394(14)70999-7 [DOI] [PubMed] [Google Scholar]
48.Foster RC, Bhatti MT, Crum OM, Lesser ER, Hodge DO, Chen JJ. Reexamining the incidence of nonarteritic anterior ischemic optic neuropathy: a Rochester epidemiology project study. J Neuroophthalmol. 2024;44(3):337-341. doi: 10.1097/WNO.0000000000002102 [DOI] [PubMed] [Google Scholar]
49.Lee JY, Park KA, Oh SY. Prevalence and incidence of nonarteritic anterior ischemic optic neuropathy in South Korea: a nationwide population-based study. Br J Ophthalmol. 2018;102(7):936-941. doi: 10.1136/bjophthalmol-2017-311140 [DOI] [PubMed] [Google Scholar]
50.Roscić V, Bojić L, Marović T. The incidence of nonarteritic ischemic optic neuropathy in the Split-Dalmatia County. Article in Croatian. Acta Med Croatica. 2009;63(2):169-172. [PubMed] [Google Scholar]
51.Xu L, Wang Y, Jonas JB. Incidence of nonarteritic anterior ischemic optic neuropathy in adult Chinese: the Beijing Eye Study. Eur J Ophthalmol. 2007;17(3):459-460. doi: 10.1177/112067210701700335 [DOI] [PubMed] [Google Scholar]
52.Lee MS, Grossman D, Arnold AC, Sloan FA. Incidence of nonarteritic anterior ischemic optic neuropathy: increased risk among diabetic patients. Ophthalmology. 2011;118(5):959-963. doi: 10.1016/j.ophtha.2011.01.054 [DOI] [PMC free article] [PubMed] [Google Scholar]
53.Hamedani AG, Kim DS, Chaitanuwong P, Gonzalez LA, Moss HE, DeLott LB. Validity of administrative coding for nonarteritic ischemic optic neuropathy. J Neuroophthalmol. 2024;44(3):342-345. doi: 10.1097/WNO.0000000000002163 [DOI] [PMC free article] [PubMed] [Google Scholar]
54.Hamedani AG, De Lott LB, Deveney T, Moss HE. Validity of international classification of diseases codes for identifying neuro-ophthalmic disease in large data sets: a systematic review. J Neuroophthalmol. 2020;40(4):514-519. doi: 10.1097/WNO.0000000000000971 [DOI] [PMC free article] [PubMed] [Google Scholar]
55.Hripcsak G, Schuemie MJ, Madigan D, Ryan PB, Suchard MA. Drawing reproducible conclusions from observational clinical data with OHDSI. Yearb Med Inform. 2021;30(1):283-289. doi: 10.1055/s-0041-1726481 [DOI] [PMC free article] [PubMed] [Google Scholar]
56.Reich N, Hölscher C. The neuroprotective effects of glucagon-like peptide 1 in Alzheimer and Parkinson disease: an in-depth review. Front Neurosci. 2022;16:970925. doi: 10.3389/fnins.2022.970925 [DOI] [PMC free article] [PubMed] [Google Scholar]
57.Yang X, Qiang Q, Li N, Feng P, Wei W, Hölscher C. Neuroprotective mechanisms of glucagon-like peptide-1-based therapies in ischemic stroke: an update based on preclinical research. Front Neurol. 2022;13:844697. doi: 10.3389/fneur.2022.844697 [DOI] [PMC free article] [PubMed] [Google Scholar]
58.Niazi S, Gnesin F, Thein AS, et al. Association between glucagon-like peptide-1 receptor agonists and the risk of glaucoma in individuals with type 2 diabetes. Ophthalmology. 2024;131(9):1056-1063. doi: 10.1016/j.ophtha.2024.03.004 [DOI] [PubMed] [Google Scholar]
59.Baggio LL, Ussher JR, McLean BA, et al. The autonomic nervous system and cardiac GLP-1 receptors control heart rate in mice. Mol Metab. 2017;6(11):1339-1349. doi: 10.1016/j.molmet.2017.08.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
60.Drucker DJ, Nauck MA. The incretin system: glucagon-like peptide-1 receptor agonists and dipeptidyl peptidase-4 inhibitors in type 2 diabetes. Lancet. 2006;368(9548):1696-1705. doi: 10.1016/S0140-6736(06)69705-5 [DOI] [PubMed] [Google Scholar]
61.Kupersmith MJ, Fraser CL, Morgenstern R, Miller NR, Levin LA, Jette N; Quark207 NAION Study Group . Ophthalmic and systemic factors of acute nonarteritic anterior ischemic optic neuropathy in the Quark207 Treatment Trial. Ophthalmology. 2024;131(7):790-802. doi: 10.1016/j.ophtha.2024.01.011 [DOI] [PubMed] [Google Scholar]
62.Cai CX, Halfpenny W, Boland MV, et al. Advancing toward a common data model in ophthalmology: gap analysis of general eye examination concepts to Standard Observational Medical Outcomes Partnership (OMOP) concepts. Ophthalmol Sci. 2023;3(4):100391. doi: 10.1016/j.xops.2023.100391 [DOI] [PMC free article] [PubMed] [Google Scholar]
63.Hribar M, Baxter SL, OHDSI Eye Care and Vision Workgroup, Goetz KE. Data standardization in the OHDSI Eye Care and Vision Research workgroup [abstract]. Invest Ophthalmol Vis Sci. 2024;65(7):2420. [Google Scholar]
64.Meier JJ. Efficacy of semaglutide in a subcutaneous and an oral formulation. Front Endocrinol (Lausanne). 2021;12:645617. doi: 10.3389/fendo.2021.645617 [DOI] [PMC free article] [PubMed] [Google Scholar]
65.Liu B, Yu Y, Liu W, Deng T, Xiang D. Risk factors for nonarteritic anterior ischemic optic neuropathy: a large scale meta-analysis. Front Med (Lausanne). 2021;8:618353. doi: 10.3389/fmed.2021.618353 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials