A telemedicine system for detecting retinopathy of prematurity in high-risk infants may help streamline examination and allow ophthalmologists to concentrate on those children with the greatest need.
Take-Home Message: A telemedicine system for detecting retinopathy of prematurity in high-risk infants may help streamline examination and allow ophthalmologists to concentrate on those children with the greatest need.
By Lynda Charters; Reviewed by Graham E. Quinn, MD, MSCE
One such model tested in the e-ROP study-a multicenter, National Eye Institute-sponsored study with headquarters in Philadelphia-yielded high sensitivity and specificity values for detecting the disease in images taken by non-physician clinicians and graded by non-physician readers compared with examinations performed by ophthalmologists experienced with ROP.
“We are in the process of watching a paradigm shift from diagnostic examinations by ophthalmologists to ROP screening using digital image assessments by non-ophthalmologists,” said Graham E. Quinn, MD, MSCE, professor of ophthalmology, The Children’s Hospital of Philadelphia, University of Pennsylvania, Philadelphia.
Dr. Quinn credited this shift to the work of Anna Ells, MD, and colleagues who described referral-warranted ROP-specifically, any ROP in zone 1, any stage 3 or worse ROP, or the presence of plus disease, that was defined as “ROP of severity enough to require a diagnostic examination by an ophthalmologist to determine whether treatment is warranted.”
This description of referral-warranted ROP is consistent with the Early-Treatment ROP recommendations of type 1 plus type 2 ROP, Dr. Quinn noted.
“About 50,000 to 60,000 children at risk for ROP are born in the United States each year, but the risk is highest among the most premature-i.e., those with birth weights of less than about 3.3 pounds and those born before 30-weeks gestation,” he said.
Further, less than 10% of the infants examined require treatment.
This statistic is important considering the number of examinations these children would require. Screening for referral-warranted ROP using digital images, therefore, could substantially decrease the large number of “unnecessary” clinical examinations and identify the children who need to be treated, according to Dr. Quinn.
Though, of course, there would have to be access to ophthalmologists experienced in ROP detection and treatment for those eyes found to have referral-warranted disease, he added.
The goal of the e-ROP study (Telemedicine Approaches to Evaluating Acute-phase ROP; (http://1.usa.gov/1ExzuRg) was to determine the validity, reliability, safety, and relative cost-effectiveness of a telemedicine system for identifying infants who have ROP that is sufficiently severe ROP to require a diagnostic examination by an ophthalmologist experienced with ROP.
Primary outcome measures were the sensitivity and specificity of detecting referral-warranted ROP when grading digital retinal images using a telemedicine system and comparing those gradings with the results of diagnostic evaluations by ophthalmologists at the same session, Dr. Quinn explained.
Twelve North American centers participated in the study. Infants were enrolled and eligible for ROP examinations if birth weight was less than 1,251 grams. Of the 1,284 enrolled infants, 1,257 had ROP examinations. The mean birth weight was 864 grams, and the mean gestation age was 27 weeks. Of this group, 51% were males, 56% were Caucasian, and 29% were black. Investigators found 63.7% of these infants had some degree of ROP, and 19.4% (244 infants) had referral-warranted ROP.
When investigators looked at the rate of referral-warranted ROP by gestational age, they found “as expected, that the rate of referral-warranted ROP decreased in incidence [by] increasing gestational age,” Dr. Quinn said.
For example, at one end of the spectrum, 63% of 83 patients with a gestational age of 23 weeks or less had referral-warranted ROP, whereas 7.8% of 204 patients with a gestational age of 27 weeks had referral-warranted ROP. Only one infant with a gestational age over 29 weeks had referral-warranted ROP.
Study procedures were as follows. After enrollment, babies underwent serial diagnostic examinations by an ophthalmologist and retinal imaging (RetCam Shuttle, Clarity Medical Systems) by non-physician imagers; clinical data and images were uploaded to a central server; and trained non-physician readers graded the images. ROP-expert readers also graded a subset of images.
An array of clinical staff included certified retinal imagers; namely, ophthalmic photographers, neonatal nurse practitioners, neonatal intensive-care nurses, optical coherence tomography technicians, certified ophthalmic medical technicians, and ophthalmic technologists.
At the reading center, three trained non-physician readers, three expert readers who were ophthalmologists, and one ophthalmologist reading center adjudicator were involved.
To assess the effectiveness of the telemedicine system, all images from infants with referral-warranted ROP and from randomly selected infants without ROP were graded-i.e., a total of 5,520 image sets from 855 infants. Two trained non-physicians readers graded the images. The reading center adjudicator settled any discrepancies.
When investigators compared the diagnostic examination with the grading evaluation for one imaging session for an infant, they found a sensitivity of 90% and a negative predictive value of about 97%. When they looked at all the imaging sessions for an infant, they found a sensitivity of 97%, Dr. Quinn noted.
Finally, when investigators looked at the last session before ROP treatment in 162 infants, the sensitivity exceeded 98% with a negative predictive value of 99%.
“There is strong support for the validity of a telemedicine system to detect eyes of at-risk infants who require a diagnostic evaluation by an ophthalmologist experienced with ROP,” he said.
Graham E. Quinn, MD, MSCE
Dr. Quinn receives support from the National Eye Institute/National Institutes of Health.