A total of 3766 participants were enrolled in the MI-SIGHT program, including 1639 participants at the free clinic and 2127 participants at the FQHC (Figure 1). After excluding those participants who withdrew and therefore did not complete screening (n=52), a total of 3714 participants’ data was analyzed (1605 free clinic; 2109 FQHC). Participants were on average 54.1 years old (SD=15.9), 37.5% identified as male, 31.8% reported their race as White and 52.0% as Black or African-American, 19.0% reported Hispanic ethnicity, 20.2% did not speak English as their primary language, 42.1% had a high school education or less, 63.5% earned less than $30,000 per year, and 23.8% did not have health insurance. Self-reported health history showed that 24.3% of participants had diabetes, 45.5% hypertension, 17.2% currently smoked, 5.0% had glaucoma, and 60.8% had not had a dilated eye exam in >2 years. Many differences existed between the two sites (Table 1) and are discussed in prior publications.^18,25 The ungradable rate for imaging was 2.6% for fundus photos and 2.9% for OCT images.

Screening results revealed varying rates of pathology (Table 2). Visual impairment in the better seeing eye existed in 11.5% of participants and was significantly higher than the national prevalence rate of 2.2% (p<0.0001).²⁶ Uncorrected or under-corrected refractive error (presenting VA ≤20/50 corrected to ≥20/40 in the better seeing eye) was observed in 9.3% of participants, significantly larger than the reported national rate of 5.3% (p<0.0001).²⁷ Cataract was observed in 22.9% of participants and included both visually significant (6.3%) and non-visually significant cataracts (16.6%). The observed rate of all forms of cataract was significantly larger than the national rate of 17.0% (p<0.0001).²⁸ The observed prevalence of glaucoma/suspected glaucoma was 22.4%, which is significantly larger than national rates of glaucoma/suspected glaucoma using different populations (13.6% of Medicare beneficiaries, 0.97% of Medicaid beneficiaries, 1.66% of commercially insured beneficiaries,²⁹ and 1.62% of adults >18 years old in the US with definite glaucoma³⁰; all p<0.0001). Definite glaucoma adjudicated by glaucoma specialists using population-based data from NHANES among adults in the US >18 years old is estimated at 1.62%³². To better compare our rate of definite glaucoma to this population based-estimate most similar to our included population, we analyzed the subset of participants who screened positive for glaucoma/suspected glaucoma and returned for in-person exam and visual field and we were able to obtain their record (n=349). In this sample, 196 were classified as glaucoma suspect, 107 as definite glaucoma, and 46 as no glaucoma. Thus, in our sample, at the lowest and most stringent end of estimation, we had a 5.3% (196/3714) prevalence of glaucoma suspect and 2.9% (107/3714) prevalence of definite glaucoma, and 2.9% is still significantly higher (p<0.0001) than the 1.62% national rate of definite glaucoma reported from NHANES. The observed prevalence of diabetic retinopathy was 4.7% and was higher than national rates (3.4%).³¹ Other pathology was found in 15.1% of participants, ranging from 0.1% of participants with Hollenhorst plaque, vascular occlusions, optic disc swelling, or ptosis obscuring vision, to 1.9% of participants with hypertensive retinopathy (Supplemental Table 1).

Of the 3714 sample, a total of 943 participants had a repeat screening (410 free clinic; 533 FQHC; Figure 1). These repeat screenings were on average 1.2 years after initial screening (SD=0.3, median = 1.1, range=0.5 to 2.9). The majority of repeat screenings were in-person (81%, 766/943) and a minority were phone screenings to collect survey data (19%, 177/943). A comparison of participants who had in-person repeat screening, phone repeat screening, or no repeat screening but a year had passed from their initial screening is presented in Supplemental Table 4. Participants with in-person repeat screening had significantly better presenting logMAR VA at their initial screening (p=0.0012), were more educated (p<0.0001), reported larger household income (p=0.0060), were more likely to have healthcare insurance (p=0.0091), and more likely to report their primary language was English (p=0.0060) compared to those with phone screening and/or those who had no screening. Participants with a repeat screening had a significant improvement in presenting logMAR VA of the worse eye from their initial screening (mean±SD, 0.25±0.59 and 0.21±0.52, respectively; p=0.0012; Table 3). Participants reported significant improvement in visual function (VFQ9) between the initial and repeat screenings for the overall composite score (81.1±14.1 and 86.4±12.0, respectively; p<0.0001) and all subscales (Table 3). Site stratified changes in logMAR VA and visual function between initial and repeat screenings showed average improvements, however some results had reduced statistical significance due to the reduced power from subgroup analysis (Table 3). Lastly, compared to initial screening results, repeat screening revealed the incident rate of glaucoma/suspected glaucoma was 6.4% (49 of 763).

Participants reported high satisfaction with the MI-SIGHT program (Supplemental Table 3). Overall, 99% of participants reported they were very satisfied (91.6%) or satisfied (7.7%) with their eye screening visit, 99% reported the program was very convenient (89.2%) or convenient (9.5%), and 99.2% reported they were very likely (93.2%) or likely (6.0%) to recommend the program to a family member or friend. No differences in program satisfaction were observed between sites (all p>0.05). Most participants ordered eyeglasses through the program (overall: 68% [2517 of 3714], free clinic: 60% [955 of 1605], FQHC: 74% [1562 of 2109]). Nearly all participants reported they were satisfied with their eyeglasses (98% overall, 99.0% free clinic, 97% at the FQHC). However, a small percentage reported they were dissatisfied (0.8% overall, 0.4% free clinic, 1.1% at the FQHC) or that their eyeglasses required a remake (1.3% overall, 0.6% free clinic, 1.7% at the FQHC).

In the initial cohort, 59% of participants screened negative for pathology and had follow up recommended according to age-related guidelines, which were all >1 year (2208 of 3714). Those who screened positive for pathology were considered a high-risk referral and were recommended for follow up in ≤ 1 year (41%, 1506 of 3714). At the end of study period, 72% of participants who were high-risk referrals (1084 of 1506) had 1 year elapse since follow-up recommendations were given. Information on attendance at recommended follow-up with an ophthalmologist within that year was obtained on 783 of these participants, of whom 529 (68%) attended follow-up. Assuming those who we were unable to obtain follow-up information on were non-adherent, the rate of attending recommended follow-up could be as low as 49% (529 of 1084). Follow-up attendance was also obtained on the subset of glaucoma participants included in the randomized trial (Supplemental Table 2). A total of 490 participants were included in the trial and randomized to either care navigation plus personalized education and motivational-interviewing-based health coaching (n=247) or care navigation plus standard written education (n=243). Attendance at recommended follow-up was similar between groups (61% versus 59% respectively, p=0.74). No difference in attendance rates was observed when stratifying by site (77% vs 71% at the free clinic, p=0.35; 51% vs 53% at the FQHC, p=0.82), although overall adherence to recommended follow-up was higher at the free clinic than at the FQHC.

Costs of the program were assessed from both a macro perspective and a micro perspective. First year study costs were previously reported and included both non-recurring startup costs ($206,369.27) and first year costs for labor and supplies ($496,206.09), excluding research costs.¹⁸ We updated startup costs to amortize the cost of the OCT and other ophthalmology equipment over the study period ($87,354.83) and updated yearly recurrent costs to exclude rent costs and reflect the reduced costs of office supplies and clinician oversight time ($393,261.59). Therefore, from a macro perspective, the cost of the program over the 3–3.5 year study period was $1,410,318.97, with a cost per participant served of $379.73, cost per case of pathology detected of $707.28, and a cost per case of glaucoma/suspected glaucoma detected of $1,695.09. In the micro-cost analysis where we used time studies to calculate the cost of the telemedicine based glaucoma screening as if it were an add-on to the infrastructure of an existing eye clinic,²³ the clinical costs were $344,831.28 (ophthalmic technician time for the screenings cost of $289,645.64 plus ophthalmologist time to review the screenings cost of $55,185.64). The equipment and supply costs for the 3.5 years was $67,370.02. Therefore, the entire program had a cost per participant served (n=3714) of $110.99, cost per case of any pathology detected (n=1994) of $206.72, and cost per case of glaucoma/suspected glaucoma detected (n=832) of $495.43.

The MI-SIGHT program screened 3,714 people for glaucoma and other eye disease over 3.5 years. The MI-SIGHT participants, recruited from a free clinic and FQHC, had higher rates of poverty compared to poverty rates in the US overall, with 63.5% having a household income less than $30,000/year compared to 25.2% of US households earning less than $35,000 per year in 2022.³² Disparities in eye health outcomes have been well documented among people with lower socio-economic status, who are disproportionately comprised of Black, African American, Latinx, and Indigenous people.³³ Placing the glaucoma and eye disease screening program in community clinics that serve people with low income enabled people who might not otherwise access specialty eye care services to receive an eye disease screening exam and low cost, affordable eyeglasses. The program identified five times the national rate of visual impairment, 1.8 times the national rate of uncorrected refractive error, 1.3 times the national rate of cataract, 1.4 times the rate of diabetic retinopathy, and 1.8 times the national rate of definite glaucoma. The observed high rate of glaucoma/suspected glaucoma of 22.4% is similar to the 25% screen positive rate observed in all three SIGHT studies combined at community locations in New York City, rural Alabama, and in the Michigan-SIGHT program. With such a high screen-positive rate for glaucoma/suspected glaucoma, the positive predictive value for bringing a SIGHT patient in for further ophthalmic examination was calculated at 86% and the false discovery rate at 14%, which would put many fewer people at risk of overtreatment and help many more people gain access to appropriate management and treatment.³⁴ People were highly satisfied with the MI-SIGHT Program and the glasses from the MI-SIGHT Program. Moreover, the MI-SIGHT Program improved outcomes. Participants vision improved because they had access to affordable glasses that were cosmetically appealing enough that participants were wearing them when they returned for their repeat screening. Participants vision-related quality of life improved, perhaps because they could see better with their affordable glasses, or perhaps because they were less worried about their eye health.

The overall rate of attendance at recommended follow-up appointments to close the loop on screening and obtain definitive care were reasonable at 68% compared to prior screening studies, where follow-up rates have ranged from 36%³⁵ to 61%³⁶ with some of the variation likely due to different program structures. Within our randomized controlled trial comparing care navigation and written information to care navigation and personalized education and motivational-interviewing based health coaching to improve attendance and recommended follow-up for those who screened positive for glaucoma, we did not find a significant difference in attendance (61% vs 60%). However, both groups did receive high levels of care navigation and both groups had reasonably high follow-up attendance rates compared to prior work evaluating follow-up rates after screening programs.^35–39

The MI-SIGHT program took place in two primary care clinics who serve people with low-income, a free clinic and an FQHC. Free clinics are highly variable in both the services they are able to provide and the volume of people they are able to care for, as these services depend on donations and volunteers for sustainability. FQHCs can bill, on a fee-for-service basis, for the medical services provided, and so increasing the volume of services rendered also increases the revenue for the FHQC, making FQHCs a potentially sustainable partner in scaling the provision of eye disease screening and optical services for medically underserved people in the United States. Currently, FQHCs are funded under Section 330 of the Public Health Service Act, which provides federal funding to offset costs of uncompensated care and aid in securing malpractice coverage for FQHCs whose mandate is then to provide primary care to medically underserved populations.⁴⁰ FQHCs predominantly provide care to people insured by Medicaid and to people who are uninsured. To make this financially sustainable, FQHCs receive a flat rate for each health care visit provided from the Prospective Payment System, a rate which is often slightly higher than the Medicaid rate for the same service in that state ($178/visit in 2021, Sherry Pace, former COO Hamilton Community Health Network, personal communication). Because telemedicine for eye care is not currently a reimbursed service at the FQHC in Michigan, no revenue was generated for the clinic from the telemedicine services provided by MI-SIGHT. If the program had been delivered by FQHC eye doctors who gave the screening results in person and confirmed these results with a slit lamp exam, they could have billed the PPS rate under CPT code G0117 for glaucoma screening for patients who 1. Had a family history of glaucoma, or 2. Identified as African American and were ≥ age 50; or 3. Identified as Hispanic or Latino and were ≥ age 65; or 4. Had diabetes or for patients who had any other ocular pathology.

Though FQHCs are often the medical home for many people insured by Medicaid and people without insurance, and provide care for 1 in 11 Americans, they are mostly not equipped to provide eye care. In 2021, only 2.6% of people accessing medical care at FQHCs received eye care, and 76% of FQHCs provided no form of eye care services, not even diabetic retinopathy screening.⁴¹ The eye care provided at FQHCs is largely concentrated in three states, with California, New York, and Massachusetts employing 56% of eye doctors at FQHCs nationally.⁴² Access to eye care is limited outside of FQHCs for people who are uninsured and even those with Medicaid. One study found that people with Medicaid, compared to those with private insurance, had 60% decreased odds of obtaining an appointment with an eye doctor when they called to schedule.⁴³ Within FQHCs, resources to help people with limited income access needed services are robust. The 1,385 FQHCs nationwide employ 5,086 full-time social workers to help with transportation and social services to support access to care.⁴² FQHCs are uniquely poised to provide direct eye disease screening, primary eye care and vision services including glasses, and help accessing recommended specialty eye care.

The MI-SIGHT program offers a paradigm for implementing eye disease screening, optical care, and eye care navigation in FQHCs that has high satisfaction, successfully detects eye disease, and resulted in improved vision and vision-related quality-of-life. It is a program that has set guidelines, based on the Veterans’ Affairs Technology Based Eye Care Services standards, for when to refer a person for eye care with a doctor.¹⁹ The protocol is based in ophthalmologists grading a standard set of history, measurements, and images and making recommendations for care, which are then communicated by ophthalmic technicians who are employed locally in the community clinic. Other eye disease screening models range from having an optometrist or ophthalmologist examine each patient with a biomicroscope to the future potential of having a medical assistant take an image of the retina and having an artificial intelligence algorithm assess for any referrable pathology.⁴⁴ Current workforce shortages projected for ophthalmologists⁴⁵ mean that a team-based approach to providing eye disease screening and optical correction is needed, where more ophthalmologist time is spent with people who have ophthalmic disease rather than spending equal amounts of time with those who do and do not have disease.

The costs of the MI-SIGHT program to detect a case of eye disease ranged from $379.73 from a macro-costing perspective to $206.72 from a micro-costing perspective. When comparing the micro-costing analyses, the cost per participant served and per case of eye disease detected was similar to the NYC-SIGHT program ($110.99 for MI-SIGHT vs $180.88 for NYC-SIGHT per participant served; $206.72 for MI-SIGHT vs $273.64 for NYC-SIGHT per case of eye disease detected).²³ NYC-SIGHT offered eye disease screening in low-income housing and low-income senior centers where people were screened by community health workers for presenting visual acuity, intraocular pressure, and had non-dilated fundus photos taken of the nerve and macula. Then, participants with vision of 20/40 or worse, IOP 23 mmHg or higher, or pathology on fundus exam were scheduled to see an on-site optometrist free-of-charge within 3 weeks. This program randomized people to receiving donated glasses free of charge or a glasses prescription, where the MI-SIGHT program leveraged a low-cost online glasses retailer to provide glasses for all participants. Both represent novel ways of bringing eye disease screening into medically underserved communities. However, currently, neither low-income housing developments nor FQHCs have the funds or resources to support this service. However, a few policy changes could make eye disease screening and optical care possible at FQHCs: 1) National Health Service Corps loan repayment would need to be opened up for eye doctors, either optometrists or ophthalmologists, to incentivize working at FQHCs, where salaries are lower and there are high no-show rates that lead to lower revenues, and there is no revenue stream from an optical shop. 2) Block grants would need to be made for the ophthalmology equipment as current fee-for-service flat rates that do not include additional coverage for ancillary testing preclude FQHCs from purchasing equipment needed for high quality screening, such as OCT. 3) Reimbursement would need to be allowed for telemedicine for eye disease screening in general, not just for diabetic retinopathy screening. 4) Medicaid fee structures would need to be augmented in some way so that participants screening positive for disease at the FQHC would be able to get definitive care from community ophthalmologists while enabling community ophthalmologists to sustain their practice. Placing eye disease screening, optical care, and care navigation in FQHCs could represent a way to mitigate the persistent disparities in eye health in the United States.