How ophthalmology researchers are applying RWD to power new insights

Across all branches of medicine, researchers are increasingly turning to real-world data (RWD) to support clinical studies and development, from study design through product launch and post-approval commercialization.

The variety of ways researchers are applying RWD to advance and accelerate ophthalmic research is both illuminating and instructive, showing not only the potential for transforming raw clinical data into actionable real-world evidence (RWE), but also examples of proven applications and strategies to improve the likelihood of successful outcomes.

Harnessing the IRIS Registry

In ophthalmology, researchers are particularly fortunate to have the vast American Academy of Ophthalmology IRIS^® Registry (Intelligent Research in Sight) as a resource, one of the largest RWD sources in existence.

The IRIS Registry is a longitudinal asset, encompassing 90 million unique patients, 962 million visits, and 12 years of data from over 15,000 clinicians across diverse electronic health record (EHR) systems.

The Registry represents data from two-thirds of ophthalmologists in the country, nearly 3,000 practices that align with population density in the US, meaning it offers a representative, high-coverage dataset.

As the data and technology partner for the American Academy of Ophthalmology, Verana Health curates the raw data from EHR notes in the IRIS Registry to generate research-ready datasets.

To generate more comprehensive views of the patient experience, Registry data can be linked to additional datasets—for example, ophthalmic images, genetic information, or claims data—using de-identified, patient-specific tokens.

“The patient journey is so complex for many of these diseases,” said Durga Borkar, MD MMCi, vitreoretinal surgeon and associate professor of ophthalmology at Duke University Eye Center. “The ability to put all this information together is really important.”

Borkar points to diseases like inherited retinal degenerations or uveitis, where patients get second or third opinions, genetic testing, images taken at multiple time points.

“Being able to marry all of those different data sources really gives a complete picture,” she said.

Learning from applications and case studies

Researchers are applying rich, targeted, disease-specific data modules to advance studies of various ophthalmology diseases. The application of RWD in various research phases illustrates the broad value and the significant potential to improve study efficiency.

Consider a case study with geographic atrophy (GA), a late-stage condition of age-related macular degeneration (AMD). Until a few years ago there were no approved treatments for GA; with the lack of incentive for clinicians to code for the disease, it was suspected that many diagnosed patients remained uncoded and therefore could not be identified based on structured data, like claims data. By extracting diagnostic information from RWD clinical notes rather than relying solely on ICD-10 codes, researchers were able to more than double the size of an identifiable GA cohort for a study.

Another disease where vital information is buried in EHR notes is thyroid eye disease (TED). There is no specific ICD-10 code for TED, making identifying patients challenging: only about 30% of Graves’ disease patients have TED, and 10% of TED patients don’t have Graves’. To gauge the severity of TED instances, a research team extracted clinical information from notes to define specific variables like proptosis measurements and Clinical Activity Scores (CAS).

RWD is also effective for regulatory and safety purposes. In one example, the FDA wanted to do a safety analysis of the most common procedures for pediatric cataracts, which is cataract surgery followed by implantation of an intraocular lens (IOL). Using RWD from the IRIS Registry, researchers were able to analyze pediatric cataract surgeries and IOL outcomes, providing critical real-world safety data for a population often excluded from traditional clinical trials.

“This study represents a pivotal step in understanding the challenges faced by a unique population of pediatric patients afflicted with cataracts,” said Malvina Eydelman, formerly with the FDA Center for Devices and Radiological Health.

RWD has also proven effective for supporting clinical trial recruitment. For a phase II study, Syneos, a clinical research organization, and EyePoint, worked with RWD provider to identify eligible sites and patients with wet AMD.

The market for study patients with AMD is crowded, according to Borkar. “There are always several studies in progress at any given site, and many of them have similar inclusion and exclusion criteria,” said Borkar. “So, it can be hard to find patients for individual trials, and sometimes people, CROs and the sponsor are behind enrollment targets.”

By applying data from the IRIS Registry and algorithmic tools, the team was able to meet enrollment targets in just 5 weeks.

“When you think about having limited resources, having a number of sites that are potentially partnering in a study, and figuring out which ones would be most high yield to go to and employ additional resources is really important,” said Borkar.

RWD continues to be important after FDA approval. In one example, the sponsor was interested in the treatment patterns and outcomes of patients who were using the intravitreal medication faricimab for diabetic macular edema (DME) during the period shortly following FDA approval—before a J code was available.

“We were able to work on this study basically right from day one, before the J-code was widely implemented, by using free text searches and creating our own code to identify patients who are on faricimab,” said Borkar. “The goal of this was to understand patient characteristics, responses to the drug, and dosing intervals, as well as treatment outcomes, including visual acuity.”

Beyond EHR notes, other types of RWD can be important to researchers, particularly images. In a case study presented at the Retina Society, researchers developed an AI machine learning-driven pipeline to identify GA using real-world fundus autofluorescence and infrared reflectance (IR) images linked to the IRIS Registry. The effort proved effective even with variable image quality from routine clinical practice.

“An algorithm like this could be really useful in screening patients for GA trials, and then also quickly identifying patients within a clinical practice that may be candidates for future GA treatments that are approved,” said Borkar.

Getting started

The variety of RWD applications illustrates how the industry is using RWD to advance drug development, whether it’s to identify patients who are candidates for certain therapeutics or to see how treatments are performing in a real-world setting.

For organizations beginning to apply RWD in clinical trial design, engaging with RWD partners early in the process is critical. This allows sponsors to better understand actual patient populations and take proactive steps to ensure that protocol requirements (e.g., specific testing intervals or equipment) align with routine clinical practice.

Inclusion criteria that on paper seem non-consequential might pose a struggle to finding patients.

“I look at protocols for a trial that we're being approached about, and it's so involved, and it requires so many invasive tests that we don't routinely do that it might be a burden, actually, for us to participate,” said Borkar. “If a protocol is so far from that norm, or if it's requiring, for example, equipment that most sites don't have, then that's going to be an added challenge.”

By engaging RWD early in the process and utilizing EHR data to strategically design study protocol, researchers improve their chances for an efficient recruitment process and trial—and ultimately improve the quality of care and life for patients.