March 2008




ROP: tackling the telemedical frontier

by Maxine Lipner Senior EyeWorld Contributing Editor



Investigators study how telemedicine meshes with ROP diagnosis

For each infant three standard images were taken of each eye Source: Michael F. Chiang, M.D.

Practitioners usually like to take a hands-on approach to ocular diagnosis. Yet new results suggest that at least when it comes to retinopathy of prematurity (ROP) it may not be necessary for ophthalmologist to be anywhere near the vicinity to determine its scope or if babies have the condition. A study published recently in Archives of Ophthalmology found that telemedicine actually had the potential improve ROP management.

Making ROP diagnosis onsite is not a coveted area for most ophthalmologists, according to lead author Michael F. Chiang, M.D., assistant professor of ophthalmology, Columbia University College of Physicians and Surgeons, New York. “Survey data has shown that a significant number of pediatric ophthalmologists and retinal specialists don’t want to do this,” he said. “The reasons they cite include things such as concern about medical legal liability, concern about poor reimbursement, and concern about the logistical difficulties of doing the exam.”

Much of the difficulty comes down to a clash of models. “Ophthalmology is really an out-patient based specialty, where we are stationed in lanes and have outpatient offices,” Dr. Chiang said. “ROP is an in-patient model, based around neo-natal ICUs.” This often means trudging equipment to the hospital and then from unit to unit. In addition, treating these young patients takes a hefty commitment. “Babies don’t go away, and if you commit to doing ROP for a nursery, for example, if you go on vacation, you may still be the one who is responsible,” Dr. Chiang said. “All of these real-world difficulties make it unattractive for ophthalmologists to want to do this.”

Taking a telemedical approach

Because ROP diagnosis is guided by a universally accepted classification system, investigators in the study thought that it might lend itself to a telemedical approach. They set out to determine whether the accuracy, reliability, and available image quality needed for the approach would make telemedicine feasible.

In the study involving 67 consecutive infants, trained neonatal nurses took wide-angle retinal images using a RetCam (Clarity Medical Systems, Pleasanton, Calif.) at two time points—between 31 to 33 weeks postmenstral age and again between 35 and 37 weeks postmenstral age. The nurses were instructed to take images that followed a standard ROP protocol. “We got one image from the central retina, one from the temporal retina, one from the nasal and then up to two additional images per eye, if the nurse felt that they added any diagnostic value,” Dr. Chiang said. These pictures were then uploaded onto the web, where they were reviewed by three pediatric retinal specialists. The three specialists first determined whether the disease was present and if so to what degree—they diagnosed patients as having no ROP, mild ROP, type 2 pre-threshold ROP, treatment-requiring ROP. They also looked at the technical quality of the images, and at whether there was sufficient retinal coverage to make an accurate diagnosis.

By the numbers

Dr. Chiang reported that results in terms of sensitivity for mild ROP, at the 31- to 33-week mark for the three graders, it ranged from 73% to 94%. When it came to detect type 2 or greater ROP, the sensitivity at this same time point ranged from 71% to 86% and the specificity, from 93% to 97%. At this same time point, the specificity numbers for detecting eyes that required treatment ranged from 94% to 100%.

By the 35- to 37-week mark, the numbers were even better. “For the mild or worse disease, the sensitivity was between 91 and 97% and the specificity between 98 and 100%,” Dr. Chiang said. “For type 2 pre-threshold or worse the sensitivity was 100% for all the graders and the specificity was between 81 and 94%.”

When investigators also considered the reliability of the telemedical approach, they found that the results were also good. At 35 to 37 weeks, they found that intergrader reliability was 0.791 to 0.889. “Different graders almost always tended to get the same answer when they looked at the same picture,” Dr. Chiang said. Likewise, when investigators had the system randomly repeat the same image, 20% of the time they found that the grader nearly always reached the same conclusion. “Particularly at 35 to 37 weeks the agreement was near perfect to perfect,” he said. “People almost always got the same answer when they looked at the same thing twice.”

In addition, graders rated technical image technical quality as “adequate” or “possibly adequate” for diagnosis in 93.3 to 100% of eyes.

Dr. Chiang sees these results as promising. “The study shows that the accuracy and also the reliability of telemedicine diagnosis are extremely high, especially at the later postmenstrual ages of 35 to 37 weeks,” he said. “Also, the perceived quality of images that were captured by the neonatal nurses was also very high, especially at these later postmenstrual ages.”

Overall, this bodes well for ROP treatment. “It really provides potential opportunities to improve the quality and also the accessibility of ROP care,” Dr. Chiang said. “We may actually end up with better care because we could potentially get the people who are the most experienced in the area to look at the images.” As long as the images are adequate, this could mean earlier and more reliable diagnosis for the baby.

Editors’ note: Dr. Chiang is an unpaid member of the scientific advisory board for Clarity Medical Systems.

Contact Information

Chiang: 212-305-9535,

ROP: tackling the telemedical frontier ROP: tackling the telemedical frontier
Ophthalmology News - EyeWorld Magazine
283 110
283 110
True, 3