Mendelian randomization (MR) has exploded across the medical literature in recent years, and ophthalmology is no exception. Once a niche analytical technique, MR is now one of the fastest-growing methods in vision science. But according to a new editorial in Translational Vision Science & Technology, this popularity comes with a warning: MR is only as strong as the methodological rigor behind it — and when misapplied, it risks misleading clinicians and cluttering the literature with weak or erroneous causal claims.
First introduced in 2003, MR leverages genetic variants as instrumental variables to determine whether an observational association reflects true causation rather than confounding. With massive GWAS datasets now publicly available, MR has become easier and faster to run than ever before. Across medicine, publications using MR jumped from 899 in 2020 to over 6,000 annually by 2024–2025. Ophthalmology mirrored this trend, rising from 17 MR papers in 2020 to more than 220 in 2025. Even TVST saw submissions climb from 7 to 44 in just one year.
The problem? Ease of use has encouraged superficial, low-quality analyses, often built on shaky assumptions or suboptimal data sources.
MR has already shaped thinking in glaucoma, AMD, diabetic retinopathy, myopia, and IOP biology, highlighting causal relationships that would be impractical — or even impossible — to test in randomized trials. But the method rests on three core assumptions: relevance, independence, and exclusion restriction. Violations can invalidate the entire analysis, yet these assumptions are notoriously hard to verify directly.
The authors of the editorial note that many published MR papers under-report sensitivity analyses, or else tend to overlook three key assumptions that need to be satisfied in order for MR to yield valid casual inference: relevance, independence, and exclusion restriction.
To protect MR’s credibility, the editorial outlines several essential principles that should guide any high-quality MR study: careful instrument selection grounded in biology or robust GWAS associations; the use of multiple analytic models (IVW, MR-Egger, weighted median, mode-based, MR-CAUSE) to test assumptions from different angles; strict adherence to reporting standards such as STROBE-MR to ensure transparency and reproducibility; cautious interpretation, integrating MR findings with RCTs, observational data, and mechanistic biology.
Without such rigor, the authors warn, MR risks becoming “yet another overstretched method vulnerable to skepticism.” The editorial closes with a clear stance: when it is used properly, MR can be a powerful tool for ophthalmologists in determining causal inference. However, only rigorous application of MR will ensure that it transforms eye research rather than misguides it.
First introduced in 2003, MR leverages genetic variants as instrumental variables to determine whether an observational association reflects true causation rather than confounding. With massive GWAS datasets now publicly available, MR has become easier and faster to run than ever before. Across medicine, publications using MR jumped from 899 in 2020 to over 6,000 annually by 2024–2025. Ophthalmology mirrored this trend, rising from 17 MR papers in 2020 to more than 220 in 2025. Even TVST saw submissions climb from 7 to 44 in just one year.
The problem? Ease of use has encouraged superficial, low-quality analyses, often built on shaky assumptions or suboptimal data sources.
MR has already shaped thinking in glaucoma, AMD, diabetic retinopathy, myopia, and IOP biology, highlighting causal relationships that would be impractical — or even impossible — to test in randomized trials. But the method rests on three core assumptions: relevance, independence, and exclusion restriction. Violations can invalidate the entire analysis, yet these assumptions are notoriously hard to verify directly.
The authors of the editorial note that many published MR papers under-report sensitivity analyses, or else tend to overlook three key assumptions that need to be satisfied in order for MR to yield valid casual inference: relevance, independence, and exclusion restriction.
To protect MR’s credibility, the editorial outlines several essential principles that should guide any high-quality MR study: careful instrument selection grounded in biology or robust GWAS associations; the use of multiple analytic models (IVW, MR-Egger, weighted median, mode-based, MR-CAUSE) to test assumptions from different angles; strict adherence to reporting standards such as STROBE-MR to ensure transparency and reproducibility; cautious interpretation, integrating MR findings with RCTs, observational data, and mechanistic biology.
Without such rigor, the authors warn, MR risks becoming “yet another overstretched method vulnerable to skepticism.” The editorial closes with a clear stance: when it is used properly, MR can be a powerful tool for ophthalmologists in determining causal inference. However, only rigorous application of MR will ensure that it transforms eye research rather than misguides it.
