The NYC-SIGHT Study is a 5-year prospective, 2:1 cluster-randomised clinical trial. The methods and cost analysis have been previously described in detail and are highlighted below.^16,17

This research was reviewed by an independent ethical review board and conforms with the principles and applicable guidelines for the protection of human subjects in biomedical research. Columbia University Irving Medical Center Institutional Review Board/Ethics Committee (#AAAR9162) approval was obtained and all aspects of the study were conducted in accordance with the Declaration of Helsinki and compliance with Health Insurance Portability and Accountability Act. Informed consent was obtained from all participants prior to enrolment and the study was registered on ClinicalTrials.gov (NCT04271709) on 2/13/20.

Ten affordable housing developments in Harlem and Washington Heights were selected by the New York City Housing Authority, which provided access to 6640 residents, with a target enrolment of 1500 people (Figure 1, Sample Frame). These developments were selected due to their close proximity to Harlem Hospital and the Columbia University Ophthalmology Department to facilitate follow-up in-office eye exam appointments. A high proportion of potentially high-risk, adults and seniors living at or below the NYC.gov poverty measure are living in these developments¹⁸ (Figure 1). These ten developm ents were considered the cluster and randomisation was conducted by the study biostatistician.

Participants living in the seven developments randomised to the Intervention Group who failed the eye health screening and needed vision correction, received complimentary eyeglasses. If they were referred to ophthalmology, they received enhanced support from patient navigators who assisted with scheduling in-office eye exam appointments over a 1-year period. Participants living in the three developments randomised to the Usual Care Group who failed the eye health screening and needed vision correction were given an eyeglasses prescription only and a list of optical shops. No enhanced support related to in-office eye exam appointments was given to the Usual Care Group. Both groups received all the basic level of eye care services provided during the eye health screening and optometric exam.

Individuals over age 40 living independently in a New York City Housing development and were willing to consent for a baseline eye health screening were recruited.^16,17 Those who met the inclusion criteria were consented over the telephone or in person, in English or Spanish, by the bilingual study coordinators. The screening and optometric exams were conducted in the community room or senior centre located within the housing development.

Recruitment took place from October 2020 to September 2022 and English and Spanish Institutional Review Board-stamped approved flyers were distributed to apartments, posted in elevators and lobbies, sent via email by the housing development, and distributed during community and food donation events, vaccination events, and health fairs.

Participants were asked over the telephone or in-person about demographics, social determinants of health, access to an eye doctor, and year of last dilated eye exam. Ocular and medical history, ocular medications, use of prescription eyeglasses, and family history of glaucoma or blindness were captured during the pre-screening assessment.

Falls risk assessment was based on the Stopping Elderly Accidents, Deaths, and Injuries algorithm.¹⁹ Participants were asked: 1) Do you worry about falling and feel unsteady when standing or walking? 2) Have you fallen in the past year? If you have fallen, how many times? 3) Were you injured from falling and visited the Emergency Department or hospitalised due to falling in the past year? Clinical conditions related to falls risk assessment were integrated into the medical history, including problems with cardiac issues, blood pressure, heart rate and/or arrhythmia, cognitive impairment, incontinence, depression, arthritis or osteoporosis, hearing problems, and foot problems.¹⁹

Visual acuity was measured by trained community health workers using the Snellen eye chart at 6m (20 feet) and measured monocularly, with correction if available. Visual acuity was converted to the logarithm of the minimum angle of resolution (log MAR) for analysis. Those with visual acuity 6/12 or worse in either eye failed the screening and were referred and scheduled for an optometric exam by the study optometrist within 3 weeks in the same location.

The intraocular pressure was checked bilaterally by the study technician, with the Ic100 iCare rebound tonometer (iCare, Helsinki, Finland), and those with intraocular pressure 23 to 29 mmHg in either eye were rechecked and the average value was recorded. If the intraocular pressure was 23–29 mmHg, the participant failed the screening and was scheduled to see the study optometrist. If the intraocular pressure was ≥30 mmHg, the participant was considered ‘fast tracked’ and referred and scheduled to see an ophthalmologist for an in-office evaluation within 2 days.^20–21

Fundus images were taken by the study ocular photographer using an auto-focus, non-mydriatic, hand-held fundus camera (Volk Pictor Prestige, Volk Optical, Mentor, OH, USA). Each eye was evaluated, and an overall reading for the worse eye was determined as either 1) normal or abnormal without significant findings, 2) abnormal with significant findings, or 3) unreadable. Those with an unreadable image were also scheduled to see the study optometrist within 3 weeks. Those with an abnormal image were contacted immediately and referred to ophthalmology for an in-office evaluation rather than the study optometrist.

Participants with an abnormal fundus image were not included in this sub-analysis since no refraction was conducted in the community and their best corrected visual acuity and IVI could not be assessed by the study optometrist.

The study optometrist obtained baseline refractive error measurements using a portable auto refractor (QuickSee Auto refractor, Plenoptika, Cambridge, MA, USA). A manifest refraction was then conducted using loose lenses and a trial frame (Goldenwall Adjustable Trial Frame Optical Trial Lens Frame PD 54–70 mm TF-BT). Near acuity was evaluated with a reading card (Rosenbaum Pocket Vision Card, Amazon) so that any add could be determined through further loose lens refraction. Best corrected near visual acuity was recorded in both eyes in Jaeger notation and refractive error was recorded in minus cylinder form. All participants were given a copy of their eyeglass prescription.

The optometrist used a portable slit lamp [(PSL, 3010-P-2000 1) Keeler USA, Malvern, PA] and direct ophthalmo-scope (Welch Allyn Panoptic 3.5 V, Skaneateles Falls, NY, USA) to perform a non-dilated ocular health evaluation. After the evaluation, participants were educated on their ocular diagnosis and referred for an in-office evaluation for a dilated eye exam, further ocular testing, treatment, and cataract surgery if warranted.

Per the International Classification of Disease, mild vision impairment: visual acuity based on the worse eye: 6/12 to 6/18 (log MAR 0.3–0.50), moderate vision impairment: visual acuity from 6/18 to 6/60 (log MAR 0.54–1.0), severe vision impairment: visual acuity from 6/60 to 3/60 (log MAR 1.0–1.3) and blindness, including visual acuity worse than 3/60, count fingers, hand motion, light perception, and no light perception (log MAR > 1.3).²²

Levels of vision impairment were determined by the presenting visual acuity, with habitual correction if eyeglasses were worn at the initial screening. Evaluations of improvable vision impairment were based on previous research papers in this field; the goal was for study participants to achieve two or more lines of improvement, which was considered a substantial vision change, that could have an impact on quality-of-life.⁹ This measurement was not an evaluation of their general eye health or optometric care, rather a simple measurable outcome of visual improvement for a community-based setting.

Participants were determined to have IVI if they presented with 6/12 or worse vision in their better seeing eye and improved by at least two Snellen lines in either eye after refraction. The statistical analysis compared participants with vision impairment who had 2 or more lines of improvement to those with vision impairment that could not be improved (6/12 or worse vision not correctable by at least 2 lines).

Demographics, social determinants of health data, as well as ocular and medical history, were evaluated to determine significance between those who had IVI and those who did not have improvable vision impairment. Refractive error was defined as myopia greater than 0.50D or hypermetropia greater than 0.75D.

Study data were collected and managed using Research Electronic Data Capture (REDCap, Vanderbilt University, Nashville, TN) software hosted at Columbia University. Statistical analyses were performed in IBM Statistical Package for Social Sciences for Windows version 25 (IBM, Armonk, NY) and R Language and Environment for Statistical Computing version 4.1.2 (R Foundation for Statistical Computing, Vienna, Austria).

Participant characteristics were summarised for the entire sample using means and standard deviations for continuous variables and frequencies and percentages for categorical variables. Outcome measures included demographic characteristics, social determinants of health, (ethnicity, race, age, sex, employment, education level, marital status, insurance), eye health screening results, and rates of suspected glaucoma and other eye diseases in the study population.

Chi-square test and two-sample t-tests were used to determine the significance between those who had IVI and those with vision impairment not improvable by at least 2 Snellen lines. A stepwise multivariate logistic regression model was constructed using the significant variables from the chi-square and two-sample t-test to identify possible predictors/factors associated with IVI 6/12 or worse using log MAR to determine odds ratios at the 95% confidence intervals. Factors included in the model were demographics, social determinants of health, and clinical characteristics. At each step, variables were added based on the alpha-to-enter significance level of 0.05 and the alpha-to-remove significance level was set at 0.1 to exclude variables in the final model.

For all analyses, p values of ≤0.05 were considered statistically significant and all tests performed were two-sided. All variables entered into the stepwise multiple logistic regression were adjusted for other variables that were also significant. All variables were accounted for as they may impact the odds of IVI.

Demographics, social determinants of health, clinical characteristics, telehealth findings, and eye health screening results of those with IVI are shown in Tables 1–3 and Figure 1. There were no statistically significant differences in the demographics between the groups.

As shown in Table 2, there were significantly higher rates of foot problems in the IVI group (p < 0.01), and higher rates of self-reported glaucoma in the vision impairment uncorrectable group (p < 0.05). While not a statistically significant difference between the groups, of note only 27.5% of participants with IVI reported that they have their own eye doctor. A total of 143 (57%) of IVI participants had not had a dilated eye exam in the past 2 years including 53 participants who could not remember their last eye exam or never had an eye exam.

Of those with IVI, self-reported ocular conditions (pre-existing) included cataract (29.1%), glaucoma (6.8%) and a family history of glaucoma (21.5%). A total of 177 participants (70.5%) with IVI wore prescription eyeglasses. Self-reported medical conditions of those with IVI included hypertension (67.7%), diabetes (30.7%), arthritis (51.0%), and asthma/COPD (19.1%).

As anticipated by the screening failure criteria, the majority of participants with IVI (94.8%) failed the community eye health screening due to having a visual acuity 6/12 or worse, while those who had vision impairment that was not improved still had high rates of screening failure (51.4%) (p < 0.00). A total of 365 participants were invited to see the on-site optometrist; 308 attended the optometric exam (adherence rate: 83%) and 286 were found to have vision impairment (Figure 1).

The mean intraocular pressure for those with vision impairment was 14.8 mmHg ± standard deviation 3.93 mmHg, taking into account the average between the right and left eyes (Table 3). Based on telehealth image data from the worse eye for those with IVI, 143 (57%) were normal, 11 (4.4%) were abnormal with no significant findings and 97 (38.6%) had an unreadable image. Those with an abnormal image (n = 286) were not included in this analysis, as they never received a refraction, therefore whether or not they had IVI could not be assessed (Figure 1, Table 3). Those with IVI were significantly more likely to have a normal fundus image result from the screening and less likely to have an unreadable image (p < 0.05) compared to those who only had 0–1 Snellen line of improvement (Table 3).

There were no significant differences between the IVI and uncorrectable vision impairment groups regarding falls history or risk, but close to half (47.4%) of those with IVI who self-selected to participate in the screening stated they worry about falling or feel unsteady when standing or walking. More than half of participants with IVI (56.2%) and 48.6% of those with uncorrectable vision impairment were invited to conduct falls assessment tests in the field during their eye health screening (Table 3).

As shown in Figure 1, the majority (87.8% 251/286) of participants examined by the optometrist and diagnosed with refractive error had at least two lines of improvement, with some participants having as much as nine Snellen lines of improvement compared to their presenting vision. A total of 59 participants had 2 lines of improvement, 120 participants had 3 to 5 lines of improvement, and 72 participants had ≥ 6 lines of improvement (Figure 1).

As shown in Table 4, all odds ratios were adjusted for other variables that were also significant. Stepwise multivariate logistic regression analysis showed that participants with visual acuity 6/12 or worse (odds ratios 16.041, 95% confidence interval 6.009 to 42.822, p = 0.000) or those with a normal fundus image (odds ratios 2.783, 95% confidence interval 1.001 to 7.740, p = 0.05) had significantly higher odds of IVI. Participants who had pre-existing glaucoma/suspect had significantly higher odds of only 0 to 1 Snellen line improvement (odds ratios 0.222, 95% confidence interval 0.071 to 0.698, p = 0.01).

The Manhattan Vision Screening and Follow-up study is one of the first cluster-randomised clinical trials that conducted community-based eye health screening and optometric exams targeting New York City residents living in affordable housing developments. The study aimed to detect vision impairment in a high-risk, underserved population of predominately older individuals of African American race and Hispanic ethnicity.

While the evaluation of the adherence to in-office eye exam appointments for those referred to ophthalmology between the intervention and the usual care group is ongoing, the authors believe that they have effectively identified and engaged a diverse population most at risk for vision impairment and provided access to and utilisation of eye care services. Results reveal high rates of refractive error and vision impairment in the study population, mostly IVI (87.8%) providing strong evidence that community-based eye health screenings are beneficial when they include an optometric exam and eyeglasses partner.

More than 50% of the participants in both groups had not had an eye exam in at least 2 years, could not remember their last dilated eye exam, or never had an eye exam, demonstrating a need to improve access to eye care in this community. The optometrist also referred 80% of those with uncorrected refractive error (all of whom had an unreadable image) and 67.3% of those with IVI for an in-office comprehensive dilated eye exam.

If a participant had any level of vision impairment and a normal fundus image, they were very likely to have vision impairment that was improvable, as shown with ≥ 2 Snellen lines of improvement. In contrast, participants who self-reported having glaucoma, which was confirmed if they were taking eye pressure-lowering medication, were less likely to have improvable vision impairment.