To the Editor We read with interest the thoughtful Invited Commentary by Browning1 on our database study of diabetic vitrectomy.2 Browning highlights the difficulties of database studies and how they might be addressed. For example, can we manually validate the electronic operation note against the patient’s medical file, to confirm that any delamination was for tractional retinal detachment? While desirable, few database studies have permission to review individual patient’s files in this way. Browning asks whether the gauge of surgery might confound our analysis? It might. Yet database studies are not designed to isolate the effect of one variable, such as port size, which occurs among many others. They are better suited to observing outcomes in toto.