Smartphone Screening for Cataract and Anterior Segment Diseases in Rural India: Comparison to In-Person Ophthalmologist Exam

Abstract

Purpose : Evaluate the diagnostic concordance of anterior segment diagnoses made by remote ophthalmologist (RO) review of community health worker (CHW)-led smartphone screenings vs in-person eye camp ophthalmologist (ECO) exams in a rural South Indian population.

Methods : Smartphone screening consisted of Snellen visual acuity (VA) measurement, anterior segment photography by CHWs using a novel low-cost hardware attachment, and asynchronous RO review and diagnosis for each eye. ECO exams consisted of VA, pen light exam, and diagnosis for each eye. Lens status diagnoses included clear crystalline lens (CCL), immature cataract (IMC), mature cataract (MC), and pseudophakia (PCIOL). Other diagnoses included refractive error/presbyopia (REP), pterygium, and corneal opacity (Figure 1). For each diagnosis, we calculated eye-level percent agreement (PA), kappa statistic (κ), sensitivity, specificity, positive predictive value, and negative predictive value comparing smartphone screening vs. ECO exams. Association of diagnostic agreement with image quality was evaluated using generalized estimating equation models.

Results : We screened N=2157 eyes of N=1101 participants across 19 eye camps in 5 regions from August 2022 to June 2023. Mean participant age was 60±12 years and 58% were female. Smartphone screening and ECO exams showed substantial or near-perfect agreement for diagnosis of CCL (PA=88%, κ=0.63), IMC (PA=85%, κ=0.69), MC (PA=96%, κ=0.67), any cataract (PA=89%, κ=0.77), and PCIOL (PA=97%, κ=0.92); moderate agreement for pterygium (PA=94%, κ=0.47) and corneal opacity (PA=99%, κ=0.56); and fair agreement for REP (PA 87%, κ=0.36). See Table 1 for detailed results. Agreement between ROs and ECOs regarding cataract diagnosis increased with image quality (OR_{poor vs acceptable} 0.69, 95% CI 0.45–1.05; OR_{optimal vs acceptable} 1.78, 95% CI 1.31–2.43).

Conclusions : Compared to in-person ECO exams, a screening platform consisting of CHW-led smartphone imaging and RO review demonstrated good performance for diagnosing cataract and lens status. Diagnostic accuracy increased with image quality. Diagnosis of refractive error using VA and smartphone photos alone was suboptimal. Diagnosis of less common conditions such as corneal opacities requires further evaluation. This novel approach shows promise in expanding access to anterior eye disease screenings in remote, low-resource settings.

This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.