Trained non-physician evaluators who studied retinal images transmitted to computer screens at a remote central reading center successfully identified newborn infants likely to require a specialized medical evaluation for retinopathy of prematurity (ROP), a leading cause of treatable blindness. Findings from a new multicenter study strengthen the case for using telemedicine to address unmet medical needs of preterm babies worldwide who cannot be initially evaluated by ophthalmologists.
"This study provides validation for a telemedicine approach to ROP screening and could help prevent thousands of kids from going blind," said lead investigator Graham E. Quinn, M.D., MSCE, a pediatric ophthalmologist at The Children's Hospital of Philadelphia who has long experience in ROP research. Quinn also is a professor of Ophthalmology in the Perelman School of Medicine at the University of Pennsylvania.
Quinn is the corresponding author and principal investigator of a study on telemedicine evaluation of ROP conducted by the e-ROP Cooperative Group, a collaboration among neonatal intensive care units in 13 North American centers. The National Eye Institute, part of the National Institutes of Health, funded the study published today in JAMA Ophthalmology.
ROP involves an abnormal growth of blood vessels in the retina that may lead to scarring, retinal detachment and, in severe cases, blindness. Some degree of ROP occurs in more than half of all infants born at 30 weeks gestation or earlier, but only five to eight percent of cases become severe enough to warrant treatment. Because early detection and prompt treatment are essential to identifying high-risk eyes, the American Academy of Ophthalmology recommends routine screening for all infants born at or before 30 weeks gestation or weighing less than 1500 grams (3.3 pounds).
Over recent years, the U.S. has seen a decline in the number of ophthalmologists who conduct ROP screening examinations. At the same time, in middle-income nations with long-standing shortages of ophthalmologists, improved survival of premature infants has expanded the numbers of babies at risk for ROP. In parts of Latin America, Asia and Eastern Europe, rates of childhood blindness caused by ROP range from 15 to 30 percent or even higher, compared to 13 percent in the U.S.
To address the public health issue of detecting potentially serious ROP, the e-ROP Cooperative Group tested the validity of a telemedicine approach by comparing evaluations by ophthalmologists with those done independently by trained non-physician image readers.
The study team analyzed results in 1,257 premature infants in neonatal intensive care units (NICUs) at 12 study centers in the U.S. and one in Canada from 2011 to 2013. On average, the babies were 13 weeks premature and all weighed less than 1251 grams (about 2.75 pounds) at birth.
The infants all received usual care—regularly scheduled diagnostic examinations by an ophthalmologist who determined whether their ROP had a severity that warranted referral for further evaluation (designated RW-ROP). In addition, NICU staff members, called certified retinal imagers, took retinal photographs of all the infants, and those images were transmitted to trained image readers at a central location at the University of Pennsylvania. The image readers, all of them non-physicians, followed a standard protocol to assess whether features of RW-ROP were present in retinal images.
The image readers were unaware of which infants had been designated by the ophthalmologists as needing referral. The two groups had broadly similar results: the image readers identified 90 percent of the infants that ophthalmologists rated as having RW-ROP. When the readers did not find RW-ROP on grading, 87 percent of the time the ophthalmologist had not noted RW-ROP on the examination.
Among the 244 babies that the ophthalmologists identified as having findings consistent with RW-ROP, 162 subsequently received treatment. Of these 162 infants, the non-physician image readers identified RW-ROP in 159 of them, meaning that 98 times out of a hundred, the eye was identified as a high-risk eye.
Quinn and colleagues pointed out several potential advantages of telemedicine screening for ROP. Non-physician imagers could perform retinal imaging more frequently than ophthalmologists, and NICU staff can implement an imaging schedule individualized to specific babies. Grading of retinal photographs could allow a more standardized approach to ROP screening, while reducing the numbers of babies needing to be examined by ophthalmologists could thus lower the costs of routine ROP screening.
Finally, remote screening could decrease the number of unnecessary patient transfers to larger nurseries with more on-site ophthalmologists. "Telemedicine potentially gives every hospital access to excellent ROP screening," said Quinn.
Quinn added that further studies needed to be done to discover whether the results are generalizable to infants not covered in this study, such as preterm babies with higher birth weights than 1251 grams. Hospital systems also would need to acquire special cameras for taking retinal images, as well as training NICU staff and establishing remote image reading centers.
Overall, said Quinn, the results are encouraging. "Although further investigations must be done before a telemedicine approach can be broadly implemented, this is a very important next step."
The Children's Hospital of Philadelphia