The COED trial was a prospective comparative trial of a dexamethasone implant combined with aflibercept against aflibercept alone in diabetic macular oedema, followed to 48 weeks.
On the outcomes that matter most, there was no difference. Central subfield thickness differed by 32.9 microns favouring combination, with a confidence interval running from -81.5 to +15.6 and a p value of 0.183. Best-corrected visual acuity differed by 3.3 letters, again not significant. Both groups improved.
What did differ was speed and burden. Oedema resolved a median 12 weeks sooner with combination therapy, and the monotherapy group needed more injections — 5.9 against 4.9 on average, an incidence rate ratio of 1.9 that did not reach significance on its own.
And it costs something. Three eyes in the combination group developed worsening cataract and two needed medical management for raised intraocular pressure. Those are the expected steroid complications and they appeared as expected.
So this is a trade to offer selectively, not a new default. A phakic patient with years of treatment ahead is a poor candidate. A patient who struggles to attend monthly injections, or one already pseudophakic, is a much better one.
- Do not expect better vision from adding the implant: acuity and thickness at 48 weeks were no different.
- Do expect faster resolution and fewer injections, which is the real argument for it.
- Weigh cataract and pressure risk explicitly; both occurred in this trial at the numbers you would predict.
- Consider it where attendance is the binding constraint rather than where response is inadequate.
- Pseudophakic eyes carry less of the downside — lens status should be part of the decision.
The statistics, in plain English
The central result here is a negative one, and it is worth reading precisely. A difference of 32.9 microns with an interval from -81.5 to +15.6 crosses zero, so the data are compatible with combination being better, worse, or the same. That is not proof the two are equivalent — the trial may simply be too small to detect a real difference — but it does mean no benefit was demonstrated. The injection-rate difference has the same problem: an incidence rate ratio of 1.9 sounds substantial, and its interval of 0.9 to 4.0 includes 1.0. The one result that reached significance is time to resolution, log-rank p 0.013, median 12 weeks sooner. So the honest summary is: faster on the one outcome powered to show it, no different on the rest.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for ophthalmology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free