Commentary on “Retrospective cohort study of ocular adverse reactions to systemic drugs: insights from a multi-country spontaneous reporting database”

Dear Editor, We read with great interest the article by Wu et al, titled “Retrospective Cohort Study of Ocular Adverse Reactions to Systemic Drugs: Insights from a Multi-Country Spontaneous Reporting Database”[1]. Our commentary was prepared in accordance with the TITAN 2025 guidelines for transparency in artificial intelligence reporting[2]. The authors conducted a comprehensive disproportionality analysis using the FAERS database, systematically identifying 141 systemic medications with positive signals for ocular adverse reactions (OARs). Their work significantly enhances our understanding of drug-induced ocular risks across therapeutic categories and provides a valuable real-world evidence base for pharmacovigilance. We commend the authors for their rigorous methodology, particularly the use of multiple disproportionality algorithms and subgroup analyses to strengthen signal reliability. The inclusion of drug-induced onset time as a quantitative metric represents a notable advance, moving beyond traditional frequency-based risk assessment toward a more dynamic understanding of OAR timelines. In light of these findings, we wish to propose two interrelated perspectives that may further contextualize the results and inform future research and clinical practice: First, the observed variation in drug-induced onset times across medication categories – from rapid-onset anti-infectives to delayed-onset endocrine agents – suggests that OAR risk is not merely a function of drug class or mechanism, but also of temporal exposure patterns and patient-specific metabolic pathways. Rather than treating onset time as a descriptive statistic, integrating it into a time-stratified risk model could help clinicians tailor monitoring schedules based on drug category, patient age, and comorbid conditions. For instance, patients on endocrine medications may benefit from extended ocular surveillance beyond the first year of therapy, whereas those on antineoplastics may require early and frequent evaluations. Second, the concept of the “disease–drug–adverse reaction triangle” implicitly underpins this study but deserves explicit emphasis in interpreting signals. For example, the strong association between insulin and OARs – even after adjusting for diabetes – highlights that drug effects may be modulated by underlying disease states, genetic factors, or concurrent medications. Future pharmacovigilance studies could adopt a network-based analytical framework that models interactions between drug classes, patient phenotypes, and temporal patterns of adverse events. Such an approach would help distinguish between drug-specific toxicity and disease-progression confounders, especially in chronic conditions like rheumatoid arthritis or multiple sclerosis. Moreover, while FAERS data are invaluable for signal detection, the absence of clinical granularity limits causal inference. We echo the authors’ call for enhanced reporting standards and suggest that prospective registries with structured ophthalmic assessments – integrated into routine chronic disease management – could capture earlier and more nuanced OAR phenotypes, such as subclinical retinal changes or asymptomatic dry eye. In summary, Wu et al[1] have provided a foundational map of systemic drug–related ocular risks. Building on their work, we advocate for a temporally informed and patient-stratified pharmacovigilance strategy that aligns monitoring protocols with drug onset profiles and individual risk factors. This approach would not only improve OAR prevention but also support more personalized and safer long-term medication use. Sincerely,