Fresh from topping this year's Ophthalmologist Power List, Professor Gus Gazzard reflects on the growing gap between innovation and evidence – and what the field must do to close it.
You've highlighted the widening gap between the pace of therapeutic innovation and our ability to test treatments robustly. What practical changes are most urgently needed?
There have always been more questions than we have the money, time, or effort to answer within the traditional randomized control trial (RCT) framework. Robust trials that really mean something are expensive and take enormous effort – and in glaucoma, where there has been a surge of innovation in surgical options, the number of comparisons we need to make has grown faster than our capacity to make them.
We need to be honest about that, and ruthlessly critical about the quality of data we do get. Too often you see meta-analyses concluding "maybe this, maybe that," because the individual studies simply aren't strong enough. If those same groups had collaborated, we might have had a definitive answer. Greater collaborative networks – clinicians and surgeons working toward shared questions rather than in silos – would make a real difference.
The other development I'd like to see is registry-based randomized controlled trials. The idea is straightforward: instead of building expensive additional infrastructure, you work within routine clinical practice. You obtain ethics approval and patient consent upfront, randomize patients where there is genuine clinical equipoise, and then let treatment proceed as normal. With modern electronic patient records, data collection can be largely automated in a governance-compliant way – and many more clinicians can participate, since they're already entering data through these systems.
My pipe dream is that within five years, a patient attending a clinic – whether in Greece or the NHS in north England – could be offered participation in such a trial as part of routine care. If they agree, they're randomized, and the data flows automatically from the electronic record. There is already movement in this direction: Cathy Sun at UCSF is working on this in the US, David Freeman and Michael Boland are exploring similar ideas within the Epic network, and Mitchell Lawler in Sydney is looking at registry-based trials within the Fight Glaucoma Blindness! registry. The potential is real, if we can bring these efforts together.
You're also exploring newer statistical tools – WIN ratios, Fragility Indices, improved visual field endpoints. How close are we to these tools becoming standard practice?
They're not standard yet, but they're gaining traction. Professor Augusto Azuara-Blanco discussed Fragility Indices at the Glaucoma Congress in Athens this year, and there's genuine excitement about their potential as a quality metric for trial reporting.
The Fragility Index gives you a shorthand measure of how robust a result really is. Many trials conclude that A is better than B – but when you look closely, the difference can be minimal or even questionable. Requiring journals to report fragility metrics alongside results, much as they require CONSORT adherence, could make trial reporting significantly more honest and transparent.
WIN ratios address a different but equally important problem. In glaucoma trials, we rely heavily on surrogate measures like intraocular pressure, while other outcomes get bundled together or overlooked entirely. A patient might have excellent pressure control but still experience blurred vision, discomfort, or significant disruption to their life – multiple surgeries, time off work. WIN ratios allow you to capture and appropriately weight multiple outcomes simultaneously. They're already being used in cardiac trials, and I'm hopeful we'll see them used as primary outcomes in ophthalmic RCTs soon. In fact, they're already being incorporated into a major randomized trial I'm involved in – though I can't share details yet.
How do you see AI reshaping both endpoints and efficiency in future trials?
It will help if we understand what it's doing. But I think it's important to be clear: we don't need AI to make many of the improvements that matter most. We could enhance outcomes tomorrow simply by implementing the knowledge we already have more consistently in routine practice. The same applies to trials – better visual field analysis and more sensitive outcome measures are available to us right now, without any AI.
That said, AI will likely help us tackle more complex questions: how best to interpret optic nerve head or retinal ganglion cell anatomy from OCT, or how to weigh visual field data against imaging outcomes. I see it as a useful tool rather than a fundamental shift. The basic structure of trials – randomization, control, rigorous analysis – isn't going to change.
Finally, what does it mean to top this year's Power List?
It feels like a tremendous accolade. I'll admit I sometimes have imposter syndrome, and recognition like this can make that worse. But I also know the list is highly regarded internationally, and I've been genuinely moved by the kind messages I've received from collaborators in the US, China, Europe, and Australia.
I hope it opens doors – to new partnerships and collaborations that might not otherwise have happened. Because ultimately, the questions we care about in glaucoma are only going to get answered through wider collaboration and stronger networks. That's what this is really about.
