Clinical trials identified intravitreal vascular endothelial growth factor inhibitors (anti-VEGF agents) have the potential to stabilise or even improve visual acuity outcomes in neovascular age-related macular degeneration (AMD), a sight-threatening disease. Real-world evidence allows us to assess whether results from randomised controlled trials can be applied to the general population. We describe the development of global registries, in particular the Fight Retinal Blindness! registry that originated in Australia, the United Kingdom AMD Electronic Medical Records User Group and the IRIS registry in the USA. Real-world observations relating to efficacy, safety and resource utilisation of intravitreal anti-VEGF therapy for neovascular AMD are then summarised. Novel observations that would have been challenging to identify in a clinical trial setting are then highlighted, including the risk of late disease reactivation, outcomes in second versus first treated eyes, and the increased risk of posterior capsular rupture during cataract surgery in patients who have received intravitreal anti-VEGF therapy. We conclude by exploring future directions in the field. This includes the development of a global consensus on real-world outcome measures to allow greater comparison of results. Real-world neovascular AMD outcome registries can be linked with other databases to determine systemic safety or genetic predictors of treatment efficacy. Machine learning offers opportunities to extract useful insights from "Big Data" often collected in these registries. Real-world registries could be used by drug regulatory authorities and industry as an alternative to more costly and time-consuming phase 4 clinical trials, potentially allowing medication costs to be based on outcomes achieved.

Aims To examine 12-month outcomes of eyes switching from intravitreal ranibizumab to aflibercept for neovascular age-related macular degeneration (nAMD). Methods Database observational study of eyes with nAMD tracked by the Fight Retinal Blindness outcome registry that received ranibizumab for at least 12 months before switching to aflibercept and followed for at least 12 months after the switch. Visual acuity (VA) recorded at 12 months after the switch was analysed using locally weighted scatterplot smoothing curves. Lesion activity was graded according to a prospectively identified definition. Main outcomes were change in VA and treatment intervals 12 months after the treatment switch. Secondary outcomes included change in activity grading, effect of duration of treatment before switching and analysis of eyes that switched back. Results A total of 384 eyes switched from ranibizumab to aflibercept after a mean duration of 39.8 months on the original treatment. The mean VA did not change from the time of switching treatment (63.4, SD 15.9 logarithm of the minimum angle of resolution letters) to 12 months later (63.3, SD 16.7). While 10% of eyes gained 10 or more letters 12 months after the switch, 13% lost the same amount. The mean number of injections decreased by around one injection in the 12 months after switching (p<0.001), with a decrease in the proportion of choroidal neovascular membrane lesions that were graded as active. Eyes that had been treated for the longest time (49 or more months) before switching had worse vision at the point of switch but neither change in VA nor treatment interval was different between groups. The small proportion (6.9%) of eyes that switched back again to ranibizumab had already lost a mean of 5.2 letters from the first switch to the switch back and continued to lose vision at a similar rate for at least 6 months. Conclusions The mean VA of eyes that switched treatments from ranibizumab to aflibercept was not different 12 months later. There was a modest increase in treatment intervals and a somewhat greater proportion of eyes that were graded as inactive after the switch.