Highlights
- •
SCREENROP screening guidelines: birth weight <1200 g or gestational age <30 weeks.
- •
Total 100% of babies with treatment-requiring ROP were captured with SCREENROP.
- •
61 neonates (7.41%) would have been saved from unnecessary screening.
- •
Unnecessary screenings increase morbidity, parental anxiety, and resource utilization.
Purpose
Reducing unnecessary retinopathy of prematurity (ROP) screenings in neonates is important for minimizing morbidity associated with such examinations including pain for the child, anxiety in parents, and conserving healthcare resources. The Seminal Canadian Recommendations for Evidence-Based Examination of Neonates for Retinopathy of Prematurity (SCREEN-ROP) guidelines recommend only screening babies with birth weight (BW) <1200 g or gestational age (GA) <30 weeks to prevent unnecessary screenings. This study aims to retrospectively test the sensitivity of SCREEN-ROP in a large urban Canadian setting for capturing babies requiring treatment for ROP.
Design
Retrospective accuracy analysis of revised diagnostic testing criteria.
Methods
All babies born at McMaster Children Hospital from July 1, 2016 to May 31, 2024 who: 1) received at least one ROP eye exam, or 2) received ROP treatment, were included. Birth weight (BW) and gestational age (GA) at birth were collected from Canadian Neonatal Network (CNN) and BORN Ontario, which stores all information on babies admitted to level 3 and level 2 Neonatal Intensive Care Units (NICUs), respectively. SCREEN-ROP criteria were applied to the cohort. Institutional ethics approval was obtained from Hamilton Integrated Research Ethics Board (REB #13216-C) and CNN (REB #04-062).
Results
Sensitivity and specificity of the model was calculated with 95% confidence intervals. Chi-squarede test was used to compare the babies captured with SCREEN-ROP criteria versus prior criteria (GA<31 weeks or birth weight <1251 g). 823 unique babies were screened from July 1, 2016 to May 31, 2024 of which 79 unique babies were treated for ROP. Applying the SCREEN-ROP criteria captured all 79 babies requiring ROP treatment while avoiding another 61 babies (7.41%) from being unnecessarily enrolled into ROP screening. Sensitivity of SCREEN-ROP criteria for capturing babies requiring ROP treatment was 100% (95% CI: 95.44% to 100%), and the specificity was 8.20% (95% CI: 6.33% to 10.41%).
Conclusions
By screening babies with BW <1200 g or GA <30 weeks, 7.41% of babies would have been saved from unnecessary ROP screenings, and all babies requiring ROP treatment would have been captured with 100% sensitivity. Further validation in other large centers and countries may allow for broader implementation of SCREEN-ROP guidelines.
INTRODUCTION
Retinopathy of prematurity (ROP) is a proliferative retinal vascular disease that affects many infants born prematurely and is the leading cause of childhood blindness in many parts of the world. While most cases self-resolve, 5-10% of these cases progress to severe ROP, which can lead to permanent blindness without treatment. In Canada, about 40-50% of premature babies develop some stage of ROP, with 6% requiring treatment. Therefore, it is crucial to screen premature babies for severe ROP so that it is detected in the early stages and promptly treat them to prevent permanent visual impairment.
According to many screening criteria, neonates fitting the criteria must be screened regularly (sometimes as often as weekly) until their retina reaches maturity. However, the screening procedure is known to be painful, with neonates demonstrating immediate pain behaviors during the procedure as well as prolonged physiological arousal after the procedure, making it vital to prevent unnecessary screening.
There are many existing models of predicting ROP such as the Weight, Insulin-like growth factor, Neonatal ROP (WINROP) algorithm, Postnatal Growth and ROP (G-ROP), and Children’s Hospital of Philadelphia (CHOP) model. These models have been validated in various parts of the world, with WINROP showing sensitivities over 80%, and G-ROP and CHOP with sensitivities over 90%. , In Canada, the current recommendation is to screen infants with gestational age (GA) <31 weeks or birth weight (BW) <1251 g. In contrast, recommendations in USA include babies with GA <30 weeks or BW ≤1500 g, and infants with BW of 1500 g to 2000 g who have an unstable clinical course. These national guidelines are limited by the small sample size and differences in populations used to develop them.
To identify and validate the most important predictors of clinically significant ROP, the Seminal Canadian Recommendations for Evidence-Based Examination of Neonates for Retinopathy of Prematurity (SCREENROP) national population-based study was conducted utilizing a nationwide Canadian database of nearly 5000 infants admitted to 32 Level 3 neonatal intensive care units (NICU) across Canada who underwent ROP screening. The study analyzed 32 prenatal and postnatal predictors of developing ROP requiring treatment with the final recommendations for screening based on only BW and GA, as more complex models that include other neonatal risk factor variables increase in complexity and reduce potential clinical implementation. The SCREENROP paper explains the various mathematical and predictive models that were developed with the final recommendations for screening based on babies born with BW <1200 g or GA <30 weeks to prevent unnecessary screenings while capturing neonates requiring treatment and with clinically significant ROP.
Reducing the number of babies enrolled in screening may raise apprehension among neonatologists and ophthalmologists who feel that babies with severe ROP may be missed, especially when dealing with neonates who fall just out of the screening criteria. Thus, these tighter guidelines must be rigorously tested and validated in large Canadian populations.
The primary objective of this study was to test the sensitivity of the SCREENROP formula in capturing the babies who require ROP treatment at one of the largest urban Canadian centers over nearly eight years.
METHODS
STUDY DESIGN
A retrospective chart review of all NICU babies born at the McMaster Children Hospital (MCH) over almost eight years (from July 1, 2016 to May 31, 2024) who met the following criteria: (1) received at least one ROP eye exam at the Level 2 or Level 3 MCH NICU, and/or (2) received ROP treatment at MCH NICU. Criteria for treatment of ROP were based on the Early Treatment Retinopathy of Prematurity guidelines.
Data was collected from the local Hamilton Canadian Neonatal Network (CNN) and BORN Ontario, which stores all clinical and demographic information on babies admitted to level 3 and level 2 NICUs, respectively. Institutional ethics was obtained from Hamilton Integrated Research Ethics Board (REB #13216-C) and CNN (REB #04-062).
OUTCOME MEASURES
For all babies meeting inclusion criteria, birth weight (grams), gestational age at birth, gestational age at the time of ROP treatment, and a unique patient identifier number were collected.
DATA COLLECTION
The above outcome measures were received electronically from the local Hamilton CNN and BORN Ontario via a password protected document. The data included a unique patient Hospital Medical Record number (HMR) that was removed once the datasets were analysed for duplicate patients. The remaining data had the HMR number removed and stored in a de-identified format.
STATISTICAL ANALYSIS
The SCREENROP criteria was retroactively applied to identify the number of babies who required treatment and were correctly captured (true positives), and the number of babies who required treatment and were missed (false negatives). These figures were used to calculate the sensitivity and specificity of the model with 95% confidence intervals. Chi-squared test was used to compare the babies captured with SCREENROP criteria (GA <30 weeks or BW <1200 g) and prior criteria (GA <31 weeks or birth weight <1251 g) that had “treated ROP” and “no ROP.” ANOVA was used to compare the mean GA and BW between “treated ROP” and “no ROP.” Post hoc tests were conducted to examine pairwise differences between the groups if the ANOVA results were significant. A p-value of less than 0.05 was considered statistically significant for both ANOVA and Chi-Squared Test.
RESULTS
A total of 1298 screenings for ROP were done from July 1, 2016 to May 31, 2024. After de-duplication for readmissions and/or transfers between level 2 and level 3 NICU care, 823 unique babies receiving ROP screening remained. Ninety babies were treated for ROP, with 79 unique babies after de-duplication for those receiving multiple treatments. Baseline characteristics are available in Table 1 and Table 2 .
| Mean | Median | Mode | Minimum | Maximum | |
|---|---|---|---|---|---|
| Birth Weight (g) | 1003 | 940 | 890 | 450 | 3700 |
| Gestational Age at birth (weeks + days) | 26.9 w + 3.0 d | 27 w + 2 d | 23 w + 6 d | 22 w + 0 d | 38 w + 6 d |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
