We are on the precipice of a demographic shift that will challenge retina specialists (and indeed, many health care providers) in a way we've never experienced before. According to US Census Bureau data, by 2030, 21% of Americans will be older than 65 years.1 By 2034, this population is projected to outnumber children for the first time in the country's history.1 An aging population brings with it inevitable age-related health challenges. Simply put: More Americans will soon need more care, more often, from roughly the same number of providers in practice today.
This is an uncomfortable reality, particularly as retina specialists are already stretched thin. To ensure that we can continue to provide the best possible care to our patients in the face of this challenge, we must take action to adapt. Fortunately, there are tools that we can employ today to help us do so, with many more to come in the near future.
Harnessing the Power of Software Solutions
Adopting new technology in eye care has historically meant investing in hardware such as optical coherence tomography (OCT), fundus photography, and other imaging and diagnostic platforms. Of course, these will continue to play a critical role, particularly the latest versions, which allow for automated, reproducible, efficient, high-quality image capture. But software that operates independently of these individual devices and facilitates the efficient gathering, analysis, storage, and sharing of data will play an increasingly important role in helping physicians make faster, more personalized decisions for each patient.
Platforms such as Harmony (Oku Digital Health), which I use in my clinic; Discovery (RetinAI), which I use as a research tool; and Zeiss Forum (Zeiss), which I’ve trained on, serve as connected clinical data platforms, integrating imaging and diagnostic information across multiple modalities and devices, regardless of manufacturer. Not only do systems like these allow for the quick and easy viewing of multiple images and data points, but they also have the capability to communicate across campuses and with referral sources using the same platform, leading to more connected care. This approach optimizes workflow and facilitates faster, more targeted referrals, removing the difficulties and time associated with trying to read a consultation report containing images that are difficult to see and interpret.
As these subscription-based software solutions continue to evolve, they will incorporate access to increasingly sophisticated artificial intelligence (AI)–assisted diagnostic capabilities, providing physicians with enhanced functionality without having to invest in new capital equipment.
AI Assistance
Aggregating multiple data points into one platform solves part of our efficiency puzzle, but analyzing all of this data can be equally cumbersome and time-consuming. This is where platforms that incorporate AI-assisted decision-making will contribute to faster analysis. If AI can reliably measure changes over time in the same patient, physicians may be able to identify the effects of a treatment or treatment interval and quickly make more targeted, personalized decisions. Additionally, instead of relying solely on data collected during the most recent visit, which is often the case currently, data collection through software, AI, and automation will provide the ability to look at not just the last 1, 2, or 3 visits, but all visits.
Some such AI solutions are already being deployed. EyeArt (Eyenuk) and LumineticsCore (Digital Diagnostics) are 2 systems currently authorized by the US Food and Drug Administration (FDA) that use AI for diabetic retinopathy screening and are intended to help optometrists and general ophthalmologists screen for referable pathology, contributing to a more efficient and effective referral process. RetInSight GmbH, part of Oku Digital Health, has developed AI algorithms that analyze OCT images for hyporeflective spaces to detect and monitor imaging biomarkers associated with diseases such as geographic atrophy (GA), neovascular age-related macular degeneration (nAMD), diabetic macular edema (DME), and retinal vein occlusion. This technology is already available in Europe, and while not yet available in the United States, it is being applied here in clinical trials, including the DAYBREAK trial (Kodiak Sciences), which evaluated the efficacy and safety of Zenkuda (tarcocimab tedromer) and tabirafusp alfa tedromer (KSI-501) compared with aflibercept 2 mg (Eylea; Regeneron) in nAMD. This trial, the positive results of which were recently reported, used RetInSight AI for all supplemental injection adjudications.
Other AI solutions may help us leverage data from our devices to track trends in patient care more broadly and thus assist with clinical decision-making based on the totality of aggregated data.
The ability to evaluate trends in imaging across a patient population could allow physicians to tie those trends to treatments that have been administered and adjust standard treatment recommendations accordingly. Software solutions that use aggregated data to measure trends and help demonstrate responses to treatment and disease progression would make possible the kind of comparative analysis that is currently very difficult for an individual physician to perform over several appointments.
Home Monitoring
Greater patient volume will require us not only to be more efficient in the clinic but also to better triage which patients need to be seen, when, and at what intervals. Software and digital solutions that patients can use at home and then share the data with their physicians will be critical tools for stratifying patients and reducing the need for clinic visits without putting patients' vision at additional risk. One example is the Scanly home OCT imaging device (Notal Vision). This technology, which enables monitoring of nAMD, became commercially available earlier this year and is being deployed for patient care. It is also being used in the Diabetic Retinopathy Clinical Research (DRCR) Retina Network Protocol AO study, in which researchers are comparing home OCT-guided pro re nata (PRN) treatment with faricimab (Vabysmo; Genentech) with the more traditional treat-and-extend approach in patients with newly diagnosed nAMD.
Of course, several factors must be considered with a home-monitoring approach. The first is appropriate patient selection. Aflibercept 8 mg (Eylea HD; Regeneron) is now FDA-approved for dosing up to every 20 weeks, but the prospect of irreversible vision loss from an adverse event associated with reactivation of a choroidal neovascular membrane naturally leads us to a treat-and-extend approach. How many patients on 20-week intervals would actually be able to extend to 40 weeks if given the opportunity? How many patients on 16-week intervals could extend to 32 weeks? Home OCT allows for a patient-guided PRN approach through which we can explore these possibilities, but for this approach to be successful, we must rely on patients to scan accurately every day. This leads to a second important consideration: technology adoption hurdles (both real and perceived).
If a patient is savvy enough to know how to scan, the process is self-guided, quick, and easy (about 45 seconds per eye), as long as they have an area in their home with a cellular connection (no need for internet or Wi-Fi). In rural areas, such as where I practice in Wisconsin, some of my patients don't have cellular coverage in their homes. Such circumstances must be taken into consideration, along with any physical limitations that could prevent a patient from scanning as directed.
A third challenge is the payer. Payers want to know whether the cost of a monthly home OCT subscription fee will be more or less than the cost of the patient seeing the physician for monthly injections. This remains an open question. We are about to submit our first claim to Medicare and expect to learn within the next few months whether home OCT will be covered.
Embracing What's to Come
To provide the best possible care for as many patients as possible—especially considering the number expected to enter the chronic retinal disease landscape in the next 10 to 20 years—we are going to have to adopt new solutions. These include moving toward a model in which algorithmic, AI-derived, and device-agnostic software solutions aggregate and analyze data to help guide clinical decision-making based on the totality of a patient's treatment journey and disease burden.
Progress can be slow. There is always some anxiety around adopting and adjusting to new technology. To be sure, there will be challenges along the way. We are used to the fixed and finite costs associated with adopting new hardware but will need to adjust to subscription models similar to those we've become familiar with outside the office (think streaming entertainment services). There will be reimbursement issues that we will have to work through. There will be onboarding time, perhaps even an extra step for certain staff. But when you consider the meaningful use we can make of this data and how it can lead to better patient outcomes, it will easily justify the additional sunk or recurring costs and the extra time spent uploading an image. Validated data supporting the use of these technologies will aid the transition. For success stories that are beginning to emerge, we can look to our colleagues in Europe, where even more of these solutions are already being used in the clinic.
In the 1980s, phacoemulsification revolutionized cataract surgery, making it one of the safest, most common surgeries in the world. In the 1990s, the ability to image the retina using OCT led to a paradigm shift in how we manage patients. Once unfamiliar, these technologies, now standards of care, have unquestionably led to better outcomes for patients. Today, we must once again look to technology to help us maximize patient outcomes and minimize risk while also tackling the uncomfortable prospect of patient volume increasing exponentially in the next few decades. The more we begin to familiarize ourselves with these new tools, the more comfortable we will all feel using them, and the faster we will be able to hit the ground running. RP
REFERENCES
1. Vespa J, Medina L, Armstrong DM. Demographic turning points for the United States: population projections for 2020 to 2060. US Census Bureau; 2020. Current Population Reports P25-1144.