Near vision data and near correction requirements from community eye health programmes in nine countries.

BACKGROUND: Recent estimates of global prevalence of uncorrected presbyopia range from 510 to 826 million. There is a shortage of primary data regarding Near Visual Impairment (NVI) magnitude. METHODS: Near visual acuity (NVA) and NVI data was collected from over 388,000 people aged 35 or over across 9 countries, within Community Eye Health programmes between January 2022 and June 2023. In Kenya (n = 34,328), dioptric power of required near correction was also recorded, and any association with age, gender or level of NVA was assessed via linear regression model. RESULTS: 146,801 of 388,939 people failed initial near vision screening (37.74%, 95% CI 37.59-37.89%), with significantly higher prevalence of NVI in Sub-Saharan Africa than South Asia. Of those with distance acuity 6/12 or better, 27.97% failed (95% CI 27.81-28.13%) with evidence of difference between genders (p < 0.001): 30.77% of women vs. 24.47% of men. The most commonly required dioptric powers of correction were +2.00D, +2.50D and +3.00D, and required power correlated with age and NVA. CONCLUSIONS: NVI remains common among Community Eye Health programme participants aged 35 and over. Data from large scale programmes such as these provide an opportunity to contribute to more accurate epidemiological estimates, and to guide future research, resource planning and intervention, ideally with improved standardisation of testing in the future.

Rapid methods for identifying barriers and solutions to improve access to community health services: a scoping review.

BACKGROUND: The advancement of universal health coverage (UHC) is largely based on identifying and addressing barriers to accessing community health services. Traditional qualitative research approaches provide excellent insights but have unfeasibly high resource requirements for most care providers. AIM: To identify, categorise, and evaluate methods that have been used to identify barriers to and/or solutions for improving access to community-based health services, grounded in engagement with affected communities, excluding approaches that take >14 days. DESIGN & SETTING: This was a scoping review. METHOD: Following Joanna Briggs Institute (JBI) guidelines, a search was undertaken using the Cochrane Library, Ovid MEDLINE, Ovid Embase, Ovid Global Health, and Google Scholar. An information specialist designed the search, and dual independent review and data charting were used. RESULTS: In total, 44 studies were included from 30 countries, reporting on 18 different clinical services. Thirty studies used self-described 'rapid' approaches; however, the majority of these did not justify what they meant by this term. Nearly half of the studies used mixed- or multi-methods and triangulation to verify early findings. All of the qualitative studies used interviews and/or focus groups, which were often supplemented with observations, document review, and mapping activities. The use of in situ snowball and convenience sampling; community members as data collectors and cultural guides; collaborative summarisation (review of findings with community members and end-users); and deductive framework analysis expedited the research processes. There were no data on costs. CONCLUSION: There are a wide range of methods that can be used to deliver timely information about barriers to access. The methods employed in the articles reviewed tended to use traditional data collection approaches in innovative ways.

Improvement studies for equitable and evidence-based innovation: an overview of the 'IM-SEEN' model.

BACKGROUND: Health inequalities are ubiquitous, and as countries seek to expand service coverage, they are at risk of exacerbating existing inequalities unless they adopt equity-focused approaches to service delivery. MAIN TEXT: Our team has developed an equity-focused continuous improvement model that reconciles prioritisation of disadvantaged groups with the expansion of service coverage. Our new approach is based on the foundations of routinely collecting sociodemographic data; identifying left-behind groups; engaging with these service users to elicit barriers and potential solutions; and then rigorously testing these solutions with pragmatic, embedded trials. This paper presents the rationale for the model, a holistic overview of how the different elements fit together, and potential applications. Future work will present findings as the model is operationalised in eye-health programmes in Botswana, India, Kenya, and Nepal. CONCLUSION: There is a real paucity of approaches for operationalising equity. By bringing a series of steps together that force programme managers to focus on groups that are being left behind, we present a model that can be used in any service delivery setting to build equity into routine practice.

Socioeconomic position and eye health outcomes: identifying inequality in rapid population-based surveys.

OBJECTIVE: Monitoring health outcomes disaggregated by socioeconomic position (SEP) is crucial to ensure no one is left behind in efforts to achieve universal health coverage. In eye health planning, rapid population surveys are most commonly implemented; these need an SEP measure that is feasible to collect within the constraints of a streamlined examination protocol. We aimed to assess whether each of four SEP measures identified inequality-an underserved group or socioeconomic gradient-in key eye health outcomes. DESIGN: Population-based cross-sectional survey. PARTICIPANTS: A subset of 4020 adults 50 years and older from a nationally representative sample of 9188 adults aged 35 years and older in The Gambia. OUTCOME MEASURES: Blindness (presenting visual acuity (PVA) <3/60), any vision impairment (VI) (PVA <6/12), cataract surgical coverage (CSC) and effective cataract surgical coverage (eCSC) at two operable cataract thresholds (<6/12 and <6/60) analysed by one objective asset-based measure (EquityTool) and three subjective measures of relative SEP (a self-reported economic ladder question and self-reported household food adequacy and income sufficiency). RESULTS: Subjective household food adequacy and income sufficiency demonstrated a socioeconomic gradient (queuing pattern) in point estimates of any VI and CSC and eCSC at both operable cataract thresholds. Any VI, CSC <6/60 and eCSC <6/60 were worse among people who reported inadequate household food compared with those with just adequate food. Any VI and CSC <6/60 were worse among people who reported not enough household income compared with those with just enough income. Neither the subjective economic ladder question nor the objective asset-wealth measure demonstrated any socioeconomic gradient or pattern of inequality in any of the eye health outcomes. CONCLUSION: We recommend pilot-testing self-reported food adequacy and income sufficiency as SEP variables in vision and eye health surveys in other locations, including assessing the acceptability, reliability and repeatability of each question.

Rapid methods for identifying barriers and solutions to improve access to community health services: a scoping review protocol.

OBJECTIVES: Low attendance rates for community health services reflect important barriers that prevent people from receiving the care they need. Services and health systems that seek to advance Universal Health Coverage need to understand and act on these factors. Formal qualitative research is the best way to elicit barriers and identify potential solutions, however traditional approaches take months to complete and can be very expensive. We aim to map the methods that have been used to rapidly elicit barriers to accessing community health services and identify potential solutions. METHODS AND ANALYSIS: We will search MEDLINE, Embase, the Cochrane Library and Global Health for empirical studies that use rapid methods (<14 days) to elicit barriers and potential solutions from intended service beneficiaries. We will exclude hospital-based and 100% remotely delivered services. We will include studies conducted in any country from 1978 to present. We will not limit by language. Two reviewers will independently perform screening and data extraction, with disagreements resolved by a third reviewer. We will tabulate the different approaches used and present data on time, skills and financial requirements for each approach, as well as the governance framework and any strengths and weaknesses presented by the study authors. We will follow Joanna Briggs Institute (JBI) scoping review guidance and report the review using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews. ETHICS AND DISSEMINATION: Ethical approval is not required. We will share our findings in the peer-reviewed literature, at conferences, and with WHO policymakers working in this space. REGISTRATION: Open Science Framework (https://osf.io/a6r2m).

Development and Validation of a Digital (Peek) Near Visual Acuity Test for Clinical Practice, Community-Based Survey, and Research.

Purpose: Unaddressed near vision impairment (NVI) affects more than 500 million people. Testing near vision is necessary to identify those in need of services. To make such testing readily accessible, we have developed and validated a new smartphone-based near visual acuity (NVA) test: Peek Near Vision (PeekNV). Methods: Two forms of the PeekNV test were developed: (1) quantitative measurement of NVA, and (2) binary screening test for presence or absence of NVI. The validity study was carried out with 483 participants in Sagarmatha Choudhary Eye Hospital, Lahan, Nepal, using a conventional Tumbling "E" Near Point Vision Chart as the reference standard. Bland-Altman limits of agreement (LoA) were used to evaluate test agreement and test-retest repeatability. NVI screening was assessed using Cohen's kappa coefficient, sensitivity, and specificity. Results: The mean difference between PeekNV and chart NVA results was 0.008 logMAR units (95% confidence interval [CI], -0.005 to 0.021) in right eye data, and the 95% LoA between PeekNV and chart testing were within 0.235 and -0.218 logMAR. As a NVI screening tool, the overall agreement between tests was 92.9% (κ = 0.85). The positive predictive value of PeekNV was 93.2% (95% CI, 89.6% to 96.9%), and the negative predictive value 92.7% (95% CI, 88.9% to 96.4%). PeekNV had a faster NVI screening time (11.6 seconds; 95% CI, 10.5 to 12.6) than the chart (14.9 seconds; 95% CI, 13.5 to 16.2; P < 0.001). Conclusions: The PeekNV smartphone-based test produces rapid NVA test results, comparable to those of an accepted NV test. Translational Relevance: PeekNV is a validated, reliable option for NV testing for use with smartphones or digital devices.

Performance and Resource Requirements of In-Person, Voice Call, and Automated Telephone-Based Socioeconomic Data Collection Modalities for Community-Based Health Programs: A Systematic Review.

Importance: Gathering data on socioeconomic status (SES) is a prerequisite for health programs that aim to improve equity. There is a lack of evidence on which approaches offer the best combination of reliability, cost, and acceptability. Objective: To compare the performance of different approaches to gathering data on SES in community health programs. Data Sources: A search of the Cochrane Library, MEDLINE, Embase, Global Health, ClinicalTrials.gov, the World Health Organization International Clinical Trials Registry Platform, and OpenGrey from 1999 to June 29, 2021, was conducted, with no language limits. Google Scholar was also searched and the reference lists of included articles were checked to identify further studies. The search was performed on June 29, 2021. Study Selection: Any empirical study design was eligible if it compared 2 or more modalities to elicit SES data from the following 3 categories: in-person, voice call, or automated telephone-based systems. Data Extraction and Synthesis: Two reviewers independently screened titles, abstracts, and full-text articles and extracted data. They also assessed the risk of bias using Cochrane tools and assessed the certainty of the evidence using the Grading of Recommendations, Assessment, Development and Evaluation approach. Findings were synthesized thematically without meta-analysis. Main Outcomes and Measures: Response rate, equivalence, time, costs, and acceptability to patients and health care professionals. Results: The searches returned 3943 records. The 11 included studies reported data on 14 036 individuals from 7 countries, collecting data on 11 socioeconomic domains using 2 or more of the following modes: in-person surveys, computer-assisted telephone interviews (CATIs), and 2 types of automated data collection: interactive voice response calls (IVRs) and web surveys. Response rates were greater than 80% for all modes except IVRs. Equivalence was high across all modes (Cohen κ > 0.5). There were insufficient data to make robust time and cost comparisons. Patients reported high levels of acceptability providing data via IVRs, web surveys, and CATIs. Conclusions and Relevance: Selecting an appropriate and cost-effective modality to elicit SES data is an important first step toward advancing equitable effective service coverage. This systematic review did not identify evidence that remote and automated data collection modes differed from human-led and in-person approaches in terms of reliability, cost, or acceptability.

Effective cataract surgical coverage in adults aged 50 years and older: estimates from population-based surveys in 55 countries.

BACKGROUND: Cataract is the leading cause of blindness globally. Effective cataract surgical coverage (eCSC) measures the number of people in a population who have been operated on for cataract, and had a good outcome, as a proportion of all people operated on or requiring surgery. Therefore, eCSC describes service access (ie, cataract surgical coverage, [CSC]) adjusted for quality. The 74th World Health Assembly endorsed a global target for eCSC of a 30-percentage point increase by 2030. To enable monitoring of progress towards this target, we analysed Rapid Assessment of Avoidable Blindness (RAAB) survey data to establish baseline estimates of eCSC and CSC. METHODS: In this secondary analysis, we used data from 148 RAAB surveys undertaken in 55 countries (2003-21) to calculate eCSC, CSC, and the relative quality gap (% difference between eCSC and CSC). Eligible studies were any version of the RAAB survey conducted since 2000 with individual participant survey data and census population data for people aged 50 years or older in the sampling area and permission from the study's principal investigator for use of data. We compared median eCSC between WHO regions and World Bank income strata and calculated the pooled risk difference and risk ratio comparing eCSC in men and women. FINDINGS: Country eCSC estimates ranged from 3·8% (95% CI 2·1-5·5) in Guinea Bissau, 2010, to 70·3% (95% CI 65·8-74·9) in Hungary, 2015, and the relative quality gap from 10·8% (CSC: 65·7%, eCSC: 58·6%) in Argentina, 2013, to 73·4% (CSC: 14·3%, eCSC: 3·8%) in Guinea Bissau, 2010. Median eCSC was highest among high-income countries (60·5% [IQR 55·6-65·4]; n=2 surveys; 2011-15) and lowest among low-income countries (14·8%; [IQR 8·3-20·7]; n=14 surveys; 2005-21). eCSC was higher in men than women (148 studies pooled risk difference 3·2% [95% CI 2·3-4·1] and pooled risk ratio of 1·20 [95% CI 1·15-1·25]). INTERPRETATION: eCSC varies widely between countries, increases with greater income level, and is higher in men. In pursuit of 2030 targets, many countries, particularly in lower-resource settings, should emphasise quality improvement before increasing access to surgery. Equity must be embedded in efforts to improve access to surgery, with a focus on underserved groups. FUNDING: Indigo Trust, Peek Vision, and Wellcome Trust.

A School Eye Health Rapid Assessment (SEHRA) planning tool: Module to survey the magnitude and nature of local needs.

BACKGROUND: Eye conditions in children can have negative consequences on visual functioning and quality of life. There is a lack of data on the magnitude of children with eye conditions who need services for effective planning of school eye health programmes. To address this, the School Eye Health Rapid Assessment (SEHRA) tool is being developed to collect data to support school eye health programme planning. METHODS: The module, 'the magnitude and nature of local needs in school children' is the first of six modules in the SEHRA tool. The module outlines a school-based cluster survey designed to determine the magnitude of eye health needs in children. This paper outlines the survey sampling strategy, and sample size calculations. RESULTS: The requirements for the SEHRA survey indicate that in regions where a larger sample size is required, or where fewer schools are recruited to the survey, confidence in the accuracy of the data will be lower. CONCLUSIONS: The SEHRA survey module 'the magnitude and nature of local needs in school children' can be applied in any context. In certain circumstances, the confidence in the survey data will be reduced.

Improving equity, efficiency and adherence to referral in Pakistan's eye health programmes: Pre- and post-pandemic onset.

Background: Over one billion people worldwide live with avoidable blindness or vision impairment. Eye Health Programmes tackle this by providing screening, primary eye care, refractive correction, and referral to hospital eye services. One point where patients can be lost in the treatment journey is adherence to hospital referral. Context: Peek Vision's software solutions have been used in Pakistan with the goal of increasing eye health programme coverage and effectiveness. This involved collaboration between health system stakeholders, international partners, local community leaders, social organizers and "Lady Health Workers". Results: From the beginning of the programmes in November 2018, to the end of December 2021, 393,759 people have been screened, 26% of whom (n = 101,236) needed refractive services or secondary eye care, and so were referred onwards to the triage centers or hospital services. Except for a short period affected heavily by COVID-19 pandemic, the programmes reached an increasing number of people over time: screening coverage improved from 774 people per month to over 28,300 people per month. Gathering and discussing data regularly with stakeholders and implementers has enabled continuous improvement to service delivery. The quality of screening and adherence to hospital visits, gender balance differences and waiting time to hospital visits were also improved. Overall attendance to hospital appointments improved in 2020 compared to 2019 from 45% (95% CI: 42-48%) to 78% (95% CI: 76-80%) in women, and from 48% (95% CI: 45-52%) to 70% (95% CI: 68-73%) in men. These patients also accessed treatment more quickly: 30-day hospital referral adherence improved from 12% in 2019 to 66% in 2020. This approach helped to utilize refractive services more efficiently, reducing false positive referrals to triage from 10.6 to 5.9%. Hospital-based services were also utilized more efficiently, as primary eye care services and refractive services were mainly delivered at the primary healthcare level. Discussion: Despite various challenges, we demonstrate how data-driven decisions can lead to health programme systems changes, including patient counseling and appointment reminders, which can effectively improve adherence to referral, allowing programmes to better meet their community's needs.

Protocol for an automated, pragmatic, embedded, adaptive randomised controlled trial: behavioural economics-informed mobile phone-based reminder messages to improve clinic attendance in a Botswanan school-based vision screening programme.

BACKGROUND: Clinic non-attendance rates are high across the African continent. Emerging evidence suggests that phone-based reminder messages could make a small but important contribution to reducing non-attendance. We will use behavioural economics principles to develop an SMS and voice reminder message to improve attendance rates in a school-based eye screening programme in Botswana. METHODS: We will test a new theory-informed SMS and voice reminder message in a national school-based eye screening programme in Botswana. The control will be the standard SMS message used to remind parents/guardians to bring their child for ophthalmic assessment. All messages will be sent twice. The primary outcome is attendance for ophthalmic assessment. We will use an automated adaptive approach, starting with a 1:1 allocation ratio. DISCUSSION: As far as we are aware, only one other study has used behavioural economics to inform the development of reminder messages to be deployed in an African healthcare setting. Our study will use an adaptive trial design, embedded in a national screening programme. Our approach can be used to trial other forms of reminder message in the future. TRIAL REGISTRATION: ISRCTN 96528723 . Registered on 5 January 2022.

Performance and resource requirements of in-person versus voice call versus automated telephone-based socioeconomic data collection modalities for community-based health programmes: a systematic review protocol.

INTRODUCTION: Gathering data on socioeconomic status (SES) is a prerequisite for any health programme that aims to assess and improve the equitable distribution of its outcomes. Many different modalities can be used to collect SES data, ranging from (1) face-to-face elicitation, to (2) telephone-administered questionnaires, to (3) automated text message-based systems. The relative costs and perceived benefits to patients and providers of these different data collection approaches is unknown. This protocol is for a systematic review that aims to compare the resource requirements, performance characteristics, and acceptability to participants and service providers of these three approaches to collect SES data from those enrolled in health programmes. METHODS AND ANALYSIS: An information specialist will conduct searches on the Cochrane Library, MEDLINE, Embase, Global Health, ClinicalTrials.gov, the WHO ICTRP and OpenGrey. All databases will be searched from 1999 to present with no language limits used. We will also search Google Scholar and check the reference lists of relevant articles for further potentially eligible studies. Any empirical study design will be eligible if it compares two or more modalities to elicit SES data from the following three; in-person, voice call, or automated phone-based systems. Two reviewers will independently screen titles, abstracts and full-text articles; and complete data extraction. For each study, we will extract data on the modality characteristics, primary outcomes (response rate and equivalence) and secondary outcomes (time, costs and acceptability to patients and providers). We will synthesise findings thematically without meta-analysis. ETHICS AND DISSEMINATION: Ethical approval is not required, as our review will include published and publicly accessible data. This review is part of a project to improve equitable access to eye care services in low-ioncome and middle-income countries. However, the findings will be useful to policy-makers and programme managers in a range of health settings and non-health settings. We will publish our findings in a peer-reviewed journal and develop an accessible summary of results for website posting and stakeholder meetings. PROSPERO REGISTRATION NUMBER: CRD42021251959.

The economics of vision impairment and its leading causes: A systematic review.

Vision impairment (VI) can have wide ranging economic impact on individuals, households, and health systems. The aim of this systematic review was to describe and summarise the costs associated with VI and its major causes. We searched MEDLINE (16 November 2019), National Health Service Economic Evaluation Database, the Database of Abstracts of Reviews of Effects and the Health Technology Assessment database (12 December 2019) for partial or full economic evaluation studies, published between 1 January 2000 and the search dates, reporting cost data for participants with VI due to an unspecified cause or one of the seven leading causes globally: cataract, uncorrected refractive error, diabetic retinopathy, glaucoma, age-related macular degeneration, corneal opacity, trachoma. The search was repeated on 20 January 2022 to identify studies published since our initial search. Included studies were quality appraised using the British Medical Journal Checklist for economic submissions adapted for cost of illness studies. Results were synthesized in a structured narrative. Of the 138 included studies, 38 reported cost estimates for VI due to an unspecified cause and 100 reported costs for one of the leading causes. These 138 studies provided 155 regional cost estimates. Fourteen studies reported global data; 103/155 (66%) regional estimates were from high-income countries. Costs were most commonly reported using a societal (n = 48) or healthcare system perspective (n = 25). Most studies included only a limited number of cost components. Large variations in methodology and reporting across studies meant cost estimates varied considerably. The average quality assessment score was 78% (range 35-100%); the most common weaknesses were the lack of sensitivity analysis and insufficient disaggregation of costs. There was substantial variation across studies in average treatment costs per patient for most conditions, including refractive error correction (range $12-$201 ppp), cataract surgery (range $54-$3654 ppp), glaucoma (range $351-$1354 ppp) and AMD (range $2209-$7524 ppp). Future cost estimates of the economic burden of VI and its major causes will be improved by the development and adoption of a reference case for eye health. This could then be used in regular studies, particularly in countries with data gaps, including low- and middle-income countries in Asia, Eastern Europe, Oceania, Latin America and sub-Saharan Africa.

Delivering Refractive Care to Populations With Near and Distance Vision Impairment: 2 Novel Social Enterprise Models.

ABSTRACT: Vision impairment due to refractive error affects crucial time periods across the life course-the educational years for children and working years for adults. Refractive error is easily and safely corrected with glasses, but many potential beneficiaries remain uncorrected due to various barriers, which can be addressed with innovative service delivery models. This review describes evidence-based initiatives from 2 social enterprises, Peek Vision and VisionSpring, addressing barriers to refractive error correction in children and working adults, particularly in low-resource settings. The reach, implementation challenges, adoption, and future development of these 2 novel models are described, and research evidence of program effectiveness is presented.

The Impact of Hyperopia on Academic Performance Among Children: A Systematic Review.

PURPOSE: To assess the impact of uncorrected hyperopia and hyperopic spectacle correction on children's academic performance. DESIGN: Systematic review and meta-analysis. METHODS: We searched 9 electronic databases from inception to July 26, 2021, for studies assessing associations between hyperopia and academic performance. There were no restrictions on language, publication date, or geographic location. A quality checklist was applied. Random-effects models estimated pooled effect size as a standardized mean difference (SMD) in 4 outcome domains: cognitive skills, educational performance, reading skills, and reading speed. (PROSPERO registration: CRD-42021268972). RESULTS: Twenty-five studies (21 observational and 4 interventional) out of 3415 met the inclusion criteria. No full-scale randomized trials were identified. Meta-analyses of the 5 studies revealed a small but significant adverse effect on educational performance in uncorrected hyperopic compared to emmetropic children {SMD -0.18 [95% confidence interval (CI), -0.27 to -0.09]; P < 0.001, 4 studies} and a moderate negative effect on reading skills in uncorrected hyperopic compared to emmetropic children [SMD -0.46 (95% CI, -0.90 to -0.03); P = 0.036, 3 studies]. Reading skills were significantly worse in hyperopic than myopic children [SMD -0.29 (95% CI, -0.43 to -0.15); P < 0.001, 1 study]. Qualitative analysis on 10 (52.6%) of 19 studies excluded from meta-analysis found a significant (P < 0.05) association between uncorrected hyperopia and impaired academic performance. Two interventional studies found hyperopic spectacle correction significantly improved reading speed (P < 0.05). CONCLUSIONS: Evidence indicates that uncorrected hyperopia is associated with poor academic performance. Given the limitations of current methodologies, further research is needed to evaluate the impact on academic performance of providing hyperopic correction.

Sociodemographic characteristics of community eye screening participants: protocol for cross-sectional equity analyses in Botswana, India, Kenya, and Nepal.

Background: Attendance rates for eye clinics are low across low- and middle-income countries (LMICs) and exhibit marked sociodemographic inequalities. We aimed to quantify the association between a range of sociodemographic domains and attendance rates from vision screening in programmes launching in Botswana, India, Kenya and Nepal. Methods: We performed a literature review of international guidance on sociodemographic data collection. Once we had identified 13 core candidate domains (age, gender, place of residence, language, ethnicity/tribe/caste, religion, marital status, parent/guardian status, place of birth, education, occupation, income, wealth) we held workshops with researchers, academics, programme implementers, and programme designers in each country to tailor the domains and response options to the national context, basing our survey development on the USAID Demographic and Health Survey model questionnaire and the RAAB7 eye health survey methodology. The draft surveys were reviewed by health economists and piloted with laypeople before being finalised, translated, and back-translated for use in Botswana, Kenya, India, and Nepal. These surveys will be used to assess the distribution of eye disease among different sociodemographic groups, and to track attendance rates between groups in four major eye screening programmes. We gather data from 3,850 people in each country and use logistic regression to identify the groups that experience the worst access to community-based eye care services in each setting. We will use a secure, password protected android-based app to gather sociodemographic information. These data will be stored using state-of-the art security measures, complying with each country's data management legislation and UK law. Discussion: This low-risk, embedded, pragmatic, observational data collection will enable eye screening programme managers to accurately identify which sociodemographic groups are facing the highest systematic barriers to accessing care at any point in time. This information will be used to inform the development of service improvements to improve equity.

Eye health indicators for universal health coverage: results of a global expert prioritisation process.

INTRODUCTION: In its recent World Report on Vision, the WHO called for an updated approach to monitor eye health as part of universal health coverage (UHC). This project sought to develop a consensus among eye health experts from all world regions to produce a menu of indicators for countries to monitor eye health within UHC. METHODS: We reviewed the literature to create a long-list of indicators aligned to the conceptual framework for monitoring outlined in WHO's World Report on Vision. We recruited a panel of 72 global eye health experts (40% women) to participate in a two-round, online prioritisation exercise. Two-hundred indicators were presented in Round 1 and participants prioritised each on a 4-point Likert scale. The highest-ranked 95 were presented in Round 2 and were (1) scored against four criteria (feasible, actionable, reliable and internationally comparable) and (2) ranked according to their suitability as a 'core' indicator for collection by all countries. The top 30 indicators ranked by these two parameters were then used as the basis for the steering group to develop a final menu. RESULTS: The menu consists of 22 indicators, including 7 core indicators, that represent important concepts in eye health for 2020 and beyond, and are considered feasible, actionable, reliable and internationally comparable. CONCLUSION: We believe this list can inform the development of new national eye health monitoring frameworks, monitor progress on key challenges to eye health and be considered in broader UHC monitoring indices at national and international levels.

Automated feature-based grading and progression analysis of diabetic retinopathy.

BACKGROUND: In diabetic retinopathy (DR) screening programmes feature-based grading guidelines are used by human graders. However, recent deep learning approaches have focused on end to end learning, based on labelled data at the whole image level. Most predictions from such software offer a direct grading output without information about the retinal features responsible for the grade. In this work, we demonstrate a feature based retinal image analysis system, which aims to support flexible grading and monitor progression. METHODS: The system was evaluated against images that had been graded according to two different grading systems; The International Clinical Diabetic Retinopathy and Diabetic Macular Oedema Severity Scale and the UK's National Screening Committee guidelines. RESULTS: External evaluation on large datasets collected from three nations (Kenya, Saudi Arabia and China) was carried out. On a DR referable level, sensitivity did not vary significantly between different DR grading schemes (91.2-94.2.0%) and there were excellent specificity values above 93% in all image sets. More importantly, no cases of severe non-proliferative DR, proliferative DR or DMO were missed. CONCLUSIONS: We demonstrate the potential of an AI feature-based DR grading system that is not constrained to any specific grading scheme.