Diagnostic test accuracy of artificial intelligence in screening for referable diabetic retinopathy in real-world settings: A systematic review and meta-analysis.

Retrospective studies on artificial intelligence (AI) in screening for diabetic retinopathy (DR) have shown promising results in addressing the mismatch between the capacity to implement DR screening and increasing DR incidence. This review sought to evaluate the diagnostic test accuracy (DTA) of AI in screening for referable diabetic retinopathy (RDR) in real-world settings. We searched CENTRAL, PubMed, CINAHL, Scopus, and Web of Science on 9 February 2023. We included prospective DTA studies assessing AI against trained human graders (HGs) in screening for RDR in patients with diabetes. Two reviewers independently extracted data and assessed methodological quality against QUADAS-2 criteria. We used the hierarchical summary receiver operating characteristics (HSROC) model to pool estimates of sensitivity and specificity and, forest plots and SROC plots to visually examine heterogeneity in accuracy estimates. From our initial search results of 3899 studies, we included 15 studies comprising 17 datasets. Meta-analyses revealed a sensitivity of 95.33% (95%CI: 90.60-100%) and specificity of 92.01% (95%CI: 87.61-96.42%) for patient-level analysis (10 datasets, N = 45,785) while, for the eye-level analysis, sensitivity was 91.24% (95%CI: 79.15-100%) and specificity, 93.90% (95%CI: 90.63-97.16%) (7 datasets, N = 15,390). Subgroup analyses did not provide variations in the diagnostic accuracy of country classification and DR classification criteria. However, a moderate increase was observed in diagnostic accuracy in the primary-level healthcare settings: sensitivity of 99.35% (95%CI: 96.85-100%), specificity of 93.72% (95%CI: 88.83-98.61%) and, a minimal decrease in the tertiary-level healthcare settings: sensitivity of 94.71% (95%CI: 89.00-100%), specificity of 90.88% (95%CI: 83.22-98.53%). Sensitivity analyses did not show any variations in studies that included diabetic macular edema in the RDR definition, nor studies with ≥3 HGs. This review provides evidence, for the first time from prospective studies, for the effectiveness of AI in screening for RDR in real-world settings. The results may serve to strengthen existing guidelines to improve current practices.

Availability of refractive error correction services in selected Zambian hospitals: a cross-sectional quantitative study.

OBJECTIVE: To assess the availability of refractive error correction services (RECS) in selected Zambian hospitals. METHODS: Between October 2021 and March 2022, we conducted a cross-sectional study design among 20 eye health service providers selected from 20 public health facilities offering RECS in 20 districts from 8 provinces of Zambia. A stratified random sampling technique was used to select study participants. A questionnaire was developed based on the access to healthcare services framework and distributed via email to respondents to collect their perspectives on the availability of RECS. RESULTS: All 20 respondents returned the completed questionnaires. Most facilities met the Ministry of Health recommended equipment requirement, though tonometers were lacking in some facilities. Out of the 20 facilities, 75% reported having optometry technologists as the main staff offering eye health services; 10% had an ophthalmologist; no facility had an optometrist; none conducted school-based programmes and only 1 facility (5%) was able to dispense spectacles soon after refraction because it had a spectacle manufacturing workshop. CONCLUSION: These findings show limited availability of RECS in the 20 health facilities. They also confirm that challenges in staffing levels, insufficient equipment and low rate of spectacle dispensing negatively affect availability of these services. Furthermore, insufficient infrastructure undermines service delivery. Addressing these challenges is cardinal to improving RECS delivery and enhancing universal eye healthcare coverage in Zambia.

Developing a competency-based curriculum for eye care managers in Sub-Saharan Africa.

INTRODUCTION: There are insufficient healthcare providers to supply eye care services to the millions of people living with visual impairment in developing countries. Eye care providers may supply more efficient and effective services if supported by skilled managers. To help produce skilled managers, appropriate competencies for eye care managers in Sub-Saharan Africa were included in the curriculum for the Postgraduate Diploma in Community Eye Health course (PGDCEH) at the University of Cape Town. METHODS: The competencies were developed from information from course-development consultations with Sub-Saharan ophthalmologists, and heads and managers of eye care programs; minutes of curriculum planning meetings; and the competencies and skills required by general, eye and public health managers. The competencies were mapped onto the PGDCEH curriculum. RESULTS: The core expectations for eye care managers were in keeping with accepted generic management competencies and skills. Additional competencies (eg population/public health) were commensurate with those for medical and population-based clinical management, medical education and public health. Broader aspects of eye care (eg patient-centered care) were also incorporated. CONCLUSIONS: Competencies for eye care managers in Sub-Saharan Africa were developed using participatory and multidisciplinary approaches to enhance context-appropriateness. Identifying these ensures that the PGDCEH learning content reflects the actual skills required, provides clarity about learning outcomes, and forms the basis for student assessment. Further, job descriptions and workplace performance evaluation of graduates can be based on these. Ongoing evaluation of the PGDCEH implementation and outcomes will provide feedback to make improvements to the course.

Evaluation of the quality of informed consent in a vaccine field trial in a developing country setting.

BACKGROUND: Informed consent is an ethical and legal requirement for research involving human participants. However, few studies have evaluated the process, particularly in Africa. Participants in a case control study designed to identify correlates of immune protection against tuberculosis (TB) in South Africa. This study was in turn nested in a large TB vaccine efficacy trial. The aim of the study was to evaluate the quality of consent in the case control study, and to identify factors that may influence the quality of consent. Cross-sectional study conducted over a 4 month period. METHODS: Consent was obtained from parents of trial participants. These parents were asked to complete a questionnaire that contained questions about the key elements of informed consent (voluntary participation, confidentiality, the main risks and benefits, etc.). The recall (success in selecting the correct answers) and understanding (correctness of interpretation of statements presented) were measured. RESULTS: The majority of the 192 subjects interviewed obtained scores greater than 75% for both the recall and understanding sections. The median score for recall was 66%; interquartile range (IQR) = 55%-77% and for understanding 75% (IQR = 50%-87%). Most (79%) were aware of the risks and 64% knew that they participated voluntarily. Participants who had completed Grade 7 at school and higher were more likely (OR = 4.94; 95% CI = 1.57 - 15.55) to obtain scores greater than 75% for recall than those who did not. Participants who were consented by professional nurses who had worked for more than two years in research were also more likely (OR = 2.62; 95% CI = 1.35-5.07) to obtain such scores for recall than those who were not. CONCLUSION: Notwithstanding the constraints in a developing country, in a population with low levels of literacy and education, the quality of informed consent found in this study could be considered as building blocks for establishing acceptable standards for public health research. Education level of respondents and experience of research staff taking the consent were associated with good quality informed consent.

Determining causes of mortality in children enrolled in a vaccine field trial in a rural area in the Western Cape Province of South Africa.

AIM: A mortality surveillance system was developed to identify and document causes of death among children enrolled in a tuberculosis vaccine field trial in South Africa. The aims of this study were to describe causes of mortality in children enrolled in a phase IV trial comparing intradermal with percutaneous administration of Bacille Calmette Guerin, and to compare causes of mortality recorded on death certificates with those obtained by clinical record review combined with verbal autopsies (CR/VA). METHODS: For children who died, certified causes of death were compared with those determined by CR/VA. RESULTS: Among 11677 children enrolled, 177 deaths were notified over 4 years. The incidence rate of death was 6.8/1000 person-years. Follow-up time ranged from 0.03 to 35.3 months (median 4 months; interquartile range 1.4-8.5). The infant mortality rate was 12.5/1000 live births and the neonatal mortality was 3/1000 live births. Pneumonia, gastroenteritis and septicaemia were among top causes of mortality by both methods. 'Sudden unexplained' and 'ill-defined' causes were among top causes of mortality based on CR/VA, while tuberculosis and 'natural causes' were among top causes based on death certificates. Important underlying causes of mortality by CR/VA include HIV/AIDS, prematurity/low birth weight and malnutrition. In 47% of deaths there was agreement on immediate causes of death. This increased to 54% when 'natural causes' and 'sudden unexplained deaths' were included. CONCLUSION: In this cohort mortality was largely due to infectious diseases. While CR/VA provided additional information on most deaths, this was not always sufficient to assign specific causes of death.