Telemedicine for Adult Retinal Diseases in Kumasi, Ghana

Introduction Blindness in low- and middle-income countries (LMICs) remains a significant public health concern—not only for the associated individual suffering, but also because of its broader implications for economic productivity and societal advancement.[1] In West Africa, the burden of retinal disease appears to be increasing due to several converging trends. The prevalence of diabetes is rising, leading to an anticipated surge in diabetic retinopathy (DR).[2] Improvements in the life expectancy of individuals with sickle cell disease may similarly result in a greater prevalence of sickle cell retinopathy.[3,4] Furthermore, as populations age, the prevalence of vision loss due to glaucoma and cataract is expected to grow in the absence of large-scale, effective intervention.[5] Despite these emerging challenges, screening and treatment capacity across LMICs has not kept pace with growing disease burden. Scalable, sustainable models for early detection and treatment are urgently needed. In response, the Africa Eye Imaging Centre was established in Kumasi, Ghana in 2023 to pioneer a telemedicine-enabled framework for identifying and managing the most pressing causes of retinal blindness. Since its inception, the center has screened thousands of patients. In this report, we share our experience—highlighting both successes and ongoing challenges—with the aim of informing and inspiring generalizable solutions in similar resource-limited settings. A 3-Tiered Approach Tier 1: Point-of-care screening We established a point-of-care retinal screening program at non-ophthalmic clinics, initially focused on diabetic centers and now expanding to include primary care and hematology clinics. Full-time staff employed by Africa Eye Imaging Centre (AEIC), our non-governmental organization (NGO), are trained as retinal photographers and undergo Cybersight training courses for diabetic retinopathy. They perform preliminary assessments and counselling, categorizing images as “normal” or “abnormal.” Final interpretations are conducted remotely by the AEIC cofounders, a U.S.-based ophthalmology resident in collaboration with a Ghanaian vitreoretinal surgeon to ensure diagnostic accuracy. The retinal photographer will then call the patient to deliver a final diagnosis and recommendation. A proprietary telemedicine platform facilitates remote grading, automated report generation, patient scheduling, and the creation of clinic- and region-level reports. For patients with common diagnoses—such as diabetic retinopathy (DR) or sickle cell retinopathy (SCR)—a link and barcode are provided that direct them to an explanatory video in either English or Twi, the predominant local language. Within 24 hours of screening, patients are contacted by the photographers with their final diagnosis (both anterior and posterior segment) and management plan, which includes counseling and triage to Tier 1 (repeat imaging), Tier 2 (further diagnostics), or Tier 3 (treatment), as described in subsequent sections. Two types of non-mydriatic cameras are utilized in the program: the 3nethra Classic HD (5 units) and the Zeiss Visuscout 100 (2 units). Devices were procured through a combination of privately raised funds, research grants, or donations directly from the manufacturers philanthropic arm. For diabetic retinopathy, macula-centered and fovea-centered images are obtained. For patients with sickle cell disease, we use a ten-field protocol. If macular pathology is identified, a six-field montage is performed. Anterior segment photographs are also taken for all patients. In order to keep the program financially sustainable, patients are charged a small fee, between 30-50 Ghanaian cedis ($2-4 USD). In addition to patients with diabetes, individuals from other hospital departments are also referred for retinal photography. For example, patients with sickle cell disease from the hematology clinic, or those with suspected retinal pathology from the optometry and ophthalmology clinics, are routinely imaged. This model could be further expanded to include patients from the human immunodeficiency virus (HIV) clinic for screening of cytomegalovirus (CMV) retinitis or other retinal pathology commonly associated with HIV. Lessons learned and programmatic adaptations Throughout implementation, several key decisions were revisited based on real-world experience: 1. Photographer Staffing ○ Initial Strategy: Recruit nurses at diabetic clinics to perform photography. ○ Challenge: Nurses already had full clinical responsibilities and could not reliably add retinal photography to their workload. Additionally, frequent staff rotation across departments in Ghana necessitated continuous retraining. ○ Adaptation: We now employ dedicated, full-time retinal photographers and graders through our NGO, ensuring consistency and quality in image acquisition and interpretation. This approach increases the overhead of the program and requires high volume screenings in order to stay financially sustainable. Employee salaries were initially paid through grant funding. As volume increased, patients paid a small fee, of which proceeds were divided between the hospital and the NGO itself. 2. Camera Placement ○ Initial Strategy: Deploy cameras permanently in high-volume diabetic clinics at teaching hospitals and government hospitals for year-round screening. Initial attempts to increase volume aimed to reach all diabetic patients, using markets, churches, mosques, and private clinics to build out a large referral network. ○ Challenge: Initially, screening volume was low as we cast a wide net and built out a referral network. We shifted our approach to known diabetics already receiving medical treatment, which led us to place cameras directly in diabetic clinics as opposed to eye clinics. This increased patient compliance, emphasizing individuals already managing their diabetes. After this adjustment, we saw an uptick in screening within the first few months; however, screening rate declined markedly thereafter, only increasing again during annual follow-ups one year after camera placement. The decline was a result of each patient typically having 3 or 4 visits per year for labs and medication refills. Within several months of a camera being deployed at a clinic, most of the clinic’s patients had at least one visit, and thus retinal photographs were offered to nearly all patients within a relatively short period of time. Because follow-up imaging was often recommended at a one-year interval, we observed large periods of stagnation in imaging volume. This led to inefficient use of resources at static locations. ○ Adaptation: We adopted a rotational model, placing cameras at different sites throughout the year to match patient flow and maximize screening volume. Prior to the placement of the camera at each site, physicians and nurses are provided referral cards for patients which communicate the time windows in which ophthalmic screening will be offered at the clinic. Today, cameras are placed at the major teaching hospital (Komfo Anokye) and several government hospitals (KNUST, Kwadaso SDA, Manhyia, Suntreso) and expanding to include a rotating schedule of smaller hospitals. 3. Image Grading ○ Initial Strategy: Use artificial intelligence (AI) models to grade images and provide more instantaneous and comprehensive point-of-care results. ○ Challenge: No FDA-approved AI solution was compatible with the 3nethra Classic HD, and furthermore the high prevalence of ungradable images (e.g., due to cataracts in older diabetic patients) reduced AI utility for any off-label algorithms. Additionally, the performance of AI algorithms for incidental lesions (e.g. glaucoma, macular degeneration) is unknown. ○ Adaptation: We now rely on human graders, which has allowed for the identification of a broad range of incidental pathologies, including cataracts, glaucoma, chorioretinal scars, retinal vein occlusions, age-related macular degeneration, and rare findings such as Stargardt disease, astrocytic hamartomas, and optic nerve melanocytomas. These observations have informed a separate manuscript currently under submission, where we found that more than half of the referrals at a DR screening site were for pathologies other than DR and diabetic macular edema (DME). Tier 2: High-fidelity Diagnostic Imaging Hubs Tier 1 screening relies on 45-degree fundus photography, which serves as an effective triage tool; however, it is often insufficient for establishing a definitive diagnosis. Patients with suspected pathology may require additional imaging modalities such as optical coherence tomography (OCT) of the retinal nerve fiber layer (RNFL) for glaucoma suspects, OCT macula for diabetic macular edema (DME) or age-related macular degeneration (AMD), and Humphrey visual field testing for glaucoma staging. In cases of retinal vascular or inflammatory disease, ultra-widefield (UWF) imaging and fluorescein angiography (FA) are used to assess disease extent and activity. Anterior segment photographs and slit-lamp imaging are also employed for evaluating anterior pathology. Tier 2 imaging requires substantial capital investment. OCT and UWF devices each cost in excess of $50,000 USD, making it impractical to deploy them across multiple peripheral sites. To address this, we have adopted a hub-and-spoke model. The “spokes” consist of lower-cost 45-degree fundus cameras, which screen and triage patients with suspected disease. Patients requiring further evaluation are referred to a central, high-fidelity imaging hub equipped with advanced diagnostics, referred to as our “central site.” The Tier 2 imaging center is staffed by trained ophthalmic technicians. All acquired images are uploaded to the telemedicine platform and graded remotely by the same team of human graders responsible for Tier 1 evaluations. The workflow that takes place at our primary imaging center (hub) and enables accurate, multimodal diagnostics and is beneficial to both patients and providers. For patients, the imaging center is open six days per week and operates on a walk-in basis, allowing for flexible scheduling. The entire visit typically takes under one hour. For referring physicians, the model reduces the clinical burden of in-person evaluations while maintaining diagnostic accuracy. Multimodal imaging has been validated as a reliable method for staging retinal and optic nerve disease.[6,7] Following the diagnostic workup, patients receive a detailed report and are counseled via telephone regarding the next steps. These may include de-escalation back to Tier 1 for routine monitoring, continued Tier 2 follow-up (e.g., serial OCT macula for borderline treatment-warranted DME), or escalation to Tier 3 for therapeutic intervention. Lessons learned and programmatic adaptations 4. Referral Pricing ○ Initial Strategy: Refer patients requiring further evaluation to private practices for advanced diagnostic imaging. ○ Challenge: Many patients lacked the financial means to access private care. This was particularly true for patients screened at government hospitals. The cost of a full diagnostic workup at high-end private facilities could be up to seven times higher than the initial Tier 1 screening, making this option inaccessible for the majority of patients. Adaptation: All Tier 2 diagnostic services are now centralized under the Africa Eye Imaging Centre. Patients referred from Tier 1 sites are offered substantially reduced pricing to ensure that financial barriers do not prevent appropriate follow-up and care. Tier 3: Treatment Centers Ultimately, a screening program that does not offer treatment is an exercise in futility. We partner closely with local ophthalmologists and optometrists at both government hospitals and private facilities to ensure that all patients are not just diagnosed but also guided all the way through to appropriate treatment. For diseases with capacity to treat at the partnering hospital (e.g. cataract, pterygium, refractive error), the patient is referred within the hospital system at which they were screened. Conversely, for patients requiring subspecialist care (e.g. diabetic retinopathy), they are referred outside to the appropriate subspecialist. Patients requiring treatment are directly referred to partnering ophthalmologists and receive regular check-ins by screening staff to ensure they can attend follow-up appointments. Examples of treatment referrals include: Patients with proliferative diabetic retinopathy (PDR) are offered panretinal photocoagulation (PRP). Patients with vitreous hemorrhage from PDR or sickle cell retinopathy (SCR) can receive intravitreal bevacizumab. Patients with diabetic macular edema (DME) may undergo focal laser treatment or intravitreal bevacizumab. Patients with glaucoma are referred to a comprehensive ophthalmologist or glaucoma specialist. Patients with incidental retinal pathology (e.g., age-related macular degeneration, retinal vein occlusion) are referred to a retina specialist. Several hundred patients with legally blinding cataracts have been referred for cataract surgery. Lessons learned and programmatic adaptations 5. Treatment Pricing Initial Strategy: Refer patients to private clinics charging market rates. Challenge: Much like Tier 2 care, market-rate services are unaffordable for the average patient. Adaptation: We began aggregating patients for high-volume treatment days to reduce cost per treatment. For example, a vial of bevacizumab yields 10–20 doses and has a one-month shelf life. By batching procedures on a single treatment day, a single vial can serve many, significantly lowering the cost per injection. Working with our partner ophthalmologists, we have established regular injection days that provide affordable, vision-saving treatments. Additionally, when patients are referred to government teaching hospitals, ophthalmology residents assist in care delivery, further reducing overhead. 6. Resident Laser Clinic Initial Strategy: Engage ophthalmology residents to assist with laser treatment for diabetic retinopathy and sickle cell retinopathy. Challenge: Laser skills are not a standardized part of the residency curriculum in Ghana, and most residents initially lacked the competency to deliver high-quality PRP or focal laser independently. Adaptation: We implemented a structured laser training program for eight residents, including a hands-on instruction course and one-on-one mentorship for those most interested in providing laser care. This enabled residents to safely assist with laser treatment of screened patients and enhanced the clinical skill set of residents in the program. Survey data collected before and after the laser training program showed an overwhelming favorable response to the training. By having a resident-run clinic, treatment prices are reduced. 7. Vitrectomy Access Initial Strategy: Refer patients for surgical intervention when necessary—for macular holes, macula-involving tractional detachments, or non-clearing vitreous hemorrhages. Challenge: Vitrectomy remains out of reach financially for most patients, often costing the equivalent of an entire year’s income. Adaptation: Unfortunately, no low-cost solution currently exists (even when consumables are reused across multiple patients). We provide detailed counseling regarding the expected costs and leave the decision to the patient. The lessons learned from Tier 3 highlight two important points: Global health retina programs should first prioritize medical retina over surgical retina. PRP and anti-VEGF therapy are substantially more cost-effective and scalable than vitrectomy, which requires specialized equipment, high consumable costs, and advanced surgical skill. There is a pressing need to develop lower-cost surgical retina solutions so that vitrectomy is accessible to all, not just the wealthy, in low- and middle-income countries (LMICs). Through a 3-tier model of care, we have developed a scalable, sustainable telemedicine framework for addressing the growing burden of adult retinal disease in West Africa. Since the first sites launch in 2023, over 3000 patients have been screened with referrable pathology identified in 577 patients, including 61 patients with PDR. By combining point-of-care screening, centralized high-fidelity diagnostics, and coordinated treatment pathways, the program has expanded access to vision-saving care for thousands of patients. While challenges remain—particularly in surgical retina—our experience highlights the critical role of innovation, partnership, and adaptability in advancing equitable eye care delivery across LMICs settings. Financial support and sponsorship Nil. Conflict of interest There are no conflict of interest.