Metformin for Age-Related Macular Degeneration: Moving Beyond Observational Studies to Causal Inference Through Target Trial Emulation and Advanced Analytics

Background: Age-related macular degeneration (AMD) represents the leading cause of irreversible vision loss in elderly populations globally. While metformin has emerged as a promising candidate for AMD prevention based on multiple observational studies, the causal relationship remains uncertain due to inherent limitations of observational research designs.Objective: This comprehensive review critically evaluates the current evidence base for metformin in AMD prevention and treatment, with particular emphasis on methodological approaches that address causal inference, including target trial emulation frameworks, propensity score methods, and emerging applications of causal artificial intelligence.Methods: We conducted a systematic review of recent literature (2019-2025) focusing on studies employing advanced causal inference methodologies. Particular attention was given to the largest meta-analysis to date (2.68 million participants) and studies utilizing target trial emulation, propensity score matching, instrumental variable analysis, and causal AI approaches.Results: Recent meta-analytic evidence demonstrates a statistically significant protective association (pooled OR = 0.86, 95% CI: 0.79–0.93, p < 0.001) between metformin use and AMD development across 18 observational studies. However, substantial heterogeneity (I² = 90%) and inherent biases in observational designs—including immortal time bias, disease latency bias, and confounding by indication—limit causal interpretation. Studies employing propensity score matching and dose-response analyses reveal protective effects primarily at low-to-moderate cumulative doses (270-600g over 2 years). Critically, no adequately powered randomized controlled trial has yet definitively established causality.Conclusions: While observational evidence suggests potential benefit, the causal effect of metformin on AMD prevention remains unproven. Rigorous application of target trial emulation frameworks, coupled with advanced causal AI methodologies, offers a pathway to strengthen causal inference from existing observational data. However, definitive evidence requires prospective randomized trials specifically designed to test metformin's efficacy in non-diabetic populations at risk for AMD.