We are grateful to Chen et al. for their commendations and suggestions for progressing this field. Competing risks (events that preclude primary event occurrence) feature in almost all Kaplan–Meier analyses and hazard estimates can be susceptible to competing risk bias; but only if featured frequently and treated as censored events [1]. We can fortunately offer reassurance that the risks nominated by Chen et al. have minimal if any, impact on our results presented. First, glaucoma progression and cataract status were not censored allowing such individuals to contribute to hazard estimates. Furthermore, neither precludes the occurrence of (and thus doesn’t compete with) the primary event—VA improvement ascribable to macular oedema. That a composite event definition could obviate competing risk bias is an advantage we had not thought to emphasize and are grateful to Chen et al. for the opportunity to rectify this.

The potential contributions of glaucoma and cataracts to VA outcomes should rather be considered as time-varying covariables in the Cox regression models alongside the suggested time-independent covariables to better understand predictors of anti-VEGF outcomes. Yet our reported primary outcome—the overall likelihood of event occurrence following therapy initiation—would remain unaffected. Whilst our analyses censored at escalation to steroid injections or cataract surgery, neither precluded the primary event and were collectively rare (40 of 3277), minimising potential of bias even if they were competing events. Output from competing risk modelling can also be unintuitive and therefore not carried out [2, 3].

Log in or create a free account to read this content

Gain free access to this article, as well as selected content from this journal and more on nature.com

Access through your institution

or

Sign in or create an account
Continue with Google
Continue with ORCiD