PurposeReal-world data give different information on health-care delivery compared with randomised controlled trials. We aimed to evaluate the appropriateness of possible quality standards for intersite comparisons of outcomes of providing Aflibercept for neovascular age-related macular degeneration (nAMD) in clinical practice.Patients and methodsRetrospective data analysis from an electronic medical record. A consecutive series of treatment-naive patients initiated on aflibercept for nAMD, in the UK from March 2013 to October 2015. Age, visual acuity (VA) at baseline and 1 year, and injection episodes were remotely extracted in an anonymised format.ResultsThe mean baseline VA was 54.3 letters, ranging from 51.3 to 58.1 between different centres, in 5620 eyes taken from 12 centres. Out of these, 3360 were initiated on treatment more than a year before. The percentage with <35 letters at baseline was 19.9-3% and that with >70 letters was 24.8-10.7%. Eyes with ≥70 letters at 1 year ranged from 20.2 to 42.9% and those with <35 ranged from 4.5 to 21.6% across different sites. Injection rates in 1 year varied from 5.5 to 8.6, and data available at 1 year also varied from 82.3 to 46.4%.ConclusionsSignificant variation was found between sites attempting to provide the same therapeutic regime. For fair comparisons between sites, we recommend that both VA measures and process measures, such as injection numbers, retention rates, and discharge policies, are used. More work is required to explain the differences. Such real-world data are not generated in the same way as a randomised clinical trial, and maybe best used to help improve service provision.