1 Br J Ophthalmol Month 2023 Vol 0 No 0 Editorial Validation of the East London Retinopathy of Prematurity algorithm to detect treatment-warranted retinopathy of prematurity: a cohort study Sonia Moorthy , 1 Gillian G W Adams, 1 Graham Smith, 2 Susmito Biswas, 3 Waleed Aliyan, 4 Roshni Bhudia, 5 Aamir Saiyed, 6 Shad Husain 7 INTRODUCTION Retinopathy of prematurity (ROP), a vasoproliferative disorder of the prema- ture retina, is a leading cause of prevent- able childhood blindness. 1–3 More severe forms of ROP occur almost exclusively in infants born at <29 weeks gestational age (GA) and/or birth weight (BW) <1251g 4 and, if left undetected and untreated, can lead to devastating vision loss. 5 ROP screening criteria universally include GA and BW. 6 These demon- strate high sensitivities but are limited by low specificities. In the UK, treatment- warranted ROP (TW-ROP) develops in less than 5% of screened infants. 7 Manage- ment consists of anti-vascular endothelial growth factor (VEGF) injections 8 and/or laser. Over the past three decades, there is an increasing demand for ROP screening due to improved survival of extremely preterm infants. 9 This could be addressed through a targeted approach, including the development and application of ROP screening tools with improved specificity, without compromising sensitivity. 10 In 2009, the first ROP screening algo- rithm, Weight, IGF-1, Neonatal Reti- nopathy of Prematurity (WINROP), first incorporated GA, BW and early weight gain (EWG). 11 Since then, several other predictive models have been developed using at least these three variables. 11–22 The postnatal growth and ROP develop- ment model (G-ROP), 16 20 and a modi- fied G-ROP (G-ROP 180g) 20 have been studied in the largest number of infants to date. Both G-ROP and G-ROP 180g have sensitivities of 100% with narrow 95% CIs to detect type 1 ROP, and the poten- tial to reduce the number of infants who require screening by about a third. 20 As ethnicity is also a factor associated with the development of severe ROP, 21 we developed the East London Retinopathy of Prematurity (EL-ROP) 22 algorithm,in- corporating this variable in addition to GA, BW and EWG (EWG, defined as weight gain during the first 6 weeks of life). EL-ROP was developed using data from a multiethnic cohort of preterm infants at risk of ROP cared for in two tertiary neonatal units in East London. Here, we report the results of EL-ROP’s validation in another cohort cared for in one of the tertiary units that participated in the development model and another in Manchester. We have also examined its performance against national ROP screening guidelines, G-ROP and G-ROP 180g. METHODS Study design This was a retrospective cohort study on routinely collected data from the tertiary neonatal units of Homerton University Hospital National Health Service (HUH NHS) Foundation Trust, London and Manchester University Hospital National Health Service (MUH NHS) Founda- tion Trust. Both units serve a multiethnic population locally and serve as perinatal centres regionally within managed clinical networks of hospitals. Perinatal centres in the UK provide the bulk of intensive care services for extremely preterm infants. Annually, each unit admits 120–150 infants who are eligible for ROP screening. In addition, because of the expertise avail- able both units accept infants from other hospitals referred for ROP treatment. Infants eligible for ROP screening according to the UK Retinopathy of Prema- turity Guideline published in May 2008 were identified from medical records and databases. This guideline recommended ROP screening of infants with GA <32 weeks or BW <1501g. 23 In March 2022, the UK guidelines were updated and the GA criterion for screening was lowered to <31 weeks. 24 Infants born between 1 September 2015 and 22 October 2021 from HUH and between 1 January and 31 December 2020 from MUH were included in the study. Exclusion criteria were as follows: (A) death prior to commence- ment or completion of ROP screening, (B) unknown ROP screening outcome and (C) admission for ROP treatment only. ROP screening was performed by neonatal ROP nurse specialists using a wide-field retinal digital imaging system (RetCam, Clarity Medical Systems, Pleas- anton, California, USA) and/or indirect ophthalmoscopy performed by expert ROP ophthalmologists. The digital images were reviewed on the same day by the ophthalmologists. Screening commenced at 30–31 weeks postmenstrual age for those born with GA <27 weeks and for all others at 4–5 weeks postnatal age and continued at 1–2 weekly intervals until vascularisation had extended into zone III or regression of ROP was detected on at least two successive examinations. Staging of ROP was recorded using the then current grading system of the International Classi- fication of Retinopathy of Prematurity. 25 Treatment was undertaken in accordance with Early Treatment for Retinopathy of Prematurity Study criteria. 26 In the UK, in addition to babies with type 1 ROP some babies with type 2 ROP are treated. 7 Final ROP screening outcome was classified as treatment-warranted (TW) or non- treatment-warranted (non-TW) ROP. Data on ROP screening outcome, BW, GA and infant weight measurements in the first 6 weeks of life were collected and entered into a secure database. Incre- mental weight measurements during the first 6–7 weeks of life were available for the HUH cohort; only the cumulative 6-week weight measurement was available for the MUH cohort. Maternal ethnicity was self-reported as black or non-black (further classified as white, asian, chinese or other). For infants transferred into or 1 Paediatric Ophthalmology and Strabismus, Moorfelds Eye Hospital City Road Campus, London, UK 2 Data Analytics, Omnivide, Sydney, New South Wales, Australia 3 Paediatric Ophthalmology, Manchester Royal Eye Hospital, Manchester, UK 4 Paediatric Ophthalmology, Manchester University NHS Foundation Trust, Manchester, UK 5 Primary Care and Public Health, Imperial College London, London, UK 6 Emergency Department, Cardiff and Vale University Health Board, Cardiff, UK 7 Neonatology, Homerton University Hospital Neonatal Unit, London, UK Correspondence to Dr Sonia Moorthy, Moorfelds Eye Hospital City Road Campus, London, UK; sonia.moorthy@icloud.com on February 9, 2023 by guest. Protected by copyright. http://bjo.bmj.com/ Br J Ophthalmol: first published as 10.1136/bjo-2022-322522 on 9 February 2023. Downloaded from