Validity of a Telemedicine System for the Evaluation of Acute-Phase ROP
Purpose: At present, premature babies requiring treatment for retinopathy of prematuriy (ROP) are identified by skilled ophthalmologists who visit neonatal units to examine babies at risk, but because fewer than 10% of babies examined require treatment, this is an inefficient use of skilled and costly personnel. We report here a large, NEI-funded, multi-center clinical study undertaken to evaluate the validity, reliability, feasibility, safety and relative cost-effectiveness of an ROP telemedicine system to detect eyes of at-risk premature babies who meet referral warranted (RW-ROP) criteria (i.e. zone I ROP, stage 3 ROP, or plus disease) and require a diagnostic evaluation by an ophthalmologist.
Methods: After consent, babies with birth weight (BW)<1251g underwent serial ROP imaging of both eyes by non-physician imagers using a wide-field digital camera, in addition to diagnostic examinations. A standard 6-image set was electronically sent to a secure central server and masked double grading by trained non-physician readers with independent third grader adjudication assessed the presence of RW-ROP criteria. The validity of image evaluation was based on the sensitivity/specificity and predictive values compared to the reference standard diagnostic exam.
Results: 1284 babies (mean BW 864g, mean GA 27 wks) were enrolled from 13 clinical centers and underwent a median of 3 sessions of imaging and examination. RW-ROP was noted in 19.2% of babies. When compared to the examination at which RW-ROP was first noted in an eye, image evaluation had sensitivity of 82.3% (95% CI 77.5, 86.2) for detecting RW-ROP, specificity 89.6% (95% CI 87.3, 91.6), negative predictive value (NPV) 96% and positive PV (PPV) 65%. When both eyes of the baby are considered for the calculation as would routinely be done in a screening situation, the sensitivity for detection of RW-ROP was 90.5% (95% CI 85.7, 94.1) with specificity of 86.2% (95% CI 83.1, 89.0), NPV 97% and PPV 61%. When images just before treatment are considered, the image evaluation has a sensitivity of 98.6% (95% CI 94.7, 99.7), specificity 79.6% (95% CI 76.2, 82.6), NPV 99% and PPV 44%.
Conclusion: The results of this study provide support for the validity of remote evaluation by trained non-physician readers of digital retinal images of babies at risk for ROP. Further, non-physicians imagers can provide retinal images for grading. Additional analyses for feasibility, reliability, safety, and cost effectiveness are underway.
Funded by National Eye Institute of the National Institutes of Health, Department of Health and Human Services. U10 EY017014
clinicaltrials.gov national registry number: NCT01264276