Seeing Into the Future of Pediatric Eye Care With Gil Binenbaum, MD

Image of female child getting her eyes examined

Gil Binenbaum, MD, has his sights set on advancing children’s eye care by adopting new perspectives. Whether it’s developing new ways to determine when a child needs glasses or diagnosing serious conditions that could lead to blindness, the attending surgeon in the division of Ophthalmology at Children’s Hospital of Philadelphia has always looked at pediatric problems through different – and for many patients, life-altering – lenses.

Earlier this year, Dr. Binenbaum received the Young Investigator Award from the American Association for Pediatric Ophthalmology and Strabismus (AAPOS), a distinction given to pediatric ophthalmologists who have conducted high-quality and impactful research. Dr. Binenbaum, who is an associate professor of ophthalmology at the Perelman School of Medicine, was also appointed as the Richard Shafritz Endowed Chair in Pediatric Ophthalmology Research. These new honors not only recognize Dr. Binenbaum’s robust contributions in re-framing how we think about eye diseases like retinopathy of prematurity (ROP), but also give him the opportunity to share his novel ideas for improving ophthalmology research through big data.

Breakthroughs for Babies: Improving ROP Diagnosis in Newborns

In 2012, Dr. Binenbaum became principal investigator of the Postnatal Growth and ROP (G-ROP) Study Group: a large-scale, multicenter project funded by the National Institutes of Health and headquartered at CHOP. The group, which spans 29 institutions in the U.S. and Canada, addresses the need for more efficient screening of ROP, a disease of the developing retinal blood vessels in premature babies.

Our retinas act like camera film in the back of our eyes: They process images that are translated into meaningful information in our brains. Blood vessels in the retina don’t fully develop until a baby reaches full term. When an infant is born prematurely, these vessels can grow in the wrong direction, causing retinal damage, vision loss, and even lifelong blindness. To prevent such damage, ophthalmologists examine premature infants repeatedly, and when they suspect that the risk of retinal damage is high, they perform laser surgery to stop the blood vessels from growing any further.

However, Dr. Binenbaum said that the way that ophthalmologists decide which infants to examine is imprecise, leaving much room for improvement. Currently, these decisions are made based on a baby’s weight and gestational age at birth. When a baby’s birth weight or gestational age fall beneath certain levels, they receive eye examinations. The result: Many infants are examined, but only a small percentage actually require treatment.

“There are many babies getting exams; those exams can be stressful for the baby, and they are resource-intensive,” Dr. Binenbaum said. “We’re quite good at identifying who to treat, but less than 5 percent of those examined actually need to be treated.”

The other 95 percent, he explained, don’t develop ROP severe enough to require treatment or, in half of cases, don’t develop ROP at all. “We’re trying to come up with a better way to decide who to examine – a better way to screen.”

Over the last decade, Dr. Binenbaum and his colleagues have developed models that use a baby’s growth rate to more accurately identify which infants need examinations, rather than the static measures of birth weight and gestational age. The team’s most recent work was published in JAMA Ophthalmology in July.

The research builds upon work from investigators at Harvard Medical School and Goteborg University in Sweden who discovered that the hormone IGF-1 influences the growth of blood vessels in the eye. Too little IGF-1 inhibits vascular-endothelial growth factor (VEGF)-induced retinal blood vessel development – an important step in the development of ROP. A baby’s weight gain is a good measure of IGF-1 levels, so slow growth is a sign of low IGF-1. Using this information gives the researchers a way to predict weeks or more in advance that a baby might develop severe ROP. Using this approach, Dr. Binenbaum and his colleagues developed the Children’s Hospital of Philadelphia ROP model. In early studies, the CHOP ROP model correctly predicted all infants who developed severe ROP, and the study team reduced the number of infants who needed exams by 49 percent.

While promising, the CHOP ROP model and similar models developed elsewhere were limited by sample sizes that were too small; the models worked well but not as perfectly when tested in new groups. To address this problem, Dr. Binenbaum formed the G-ROP study group. In their first study, they retrospectively studied a cohort of 7,483 newborns who underwent ROP exams at 29 hospitals in the U.S. and Canada between 2006 and 2012.

The G-ROP study has produced the largest detailed ROP dataset ever created. They’ve used the data to develop new evidence-based birth weight, gestational age, and weight gain ROP screening criteria that can reduce the number of infants examined by one-third. Using data from a group of infants so large makes it much more likely that the criteria will test well when applied to new patients. To perform that validation, the study group has just completed a prospective 4,200 infant study, and Dr. Binenbaum presented these new screening criteria at the recent AAPOS annual meeting where he received the Young Investigator Award.

Data-Driven Discovery: Big Data on a Small Scale

Beyond the CHOP-ROP model, Dr. Binenbaum has many additional ideas for improving pediatric ophthalmology research. As recipient of the AAPOS Young Investigator Award, he delivered a lecture on how clinicians interested in research can use information in electronic medical records (EMR) to efficiently access large amounts of research data from their own practice.

“There’s a whole art and science to doing studies with huge medical databases,” Dr. Binenbaum said. “My idea was to inspire people in the audience to use a similar ‘big data’ approach to their own practice, whether they’re doing research already or are in clinical practice, seeing a lot of patients and coming up with good clinical questions; I called it ‘Big Data on a Small Scale.’” He adds that clinician-researchers may not realize that they have a treasure trove of data in their EMR to answer these questions.

In the lecture, Dr. Binenbaum shared many examples of how to apply this big data approach to questions in ophthalmology such as, “Do vaccinations cause retinal hemorrhage?” To answer this question, Dr. Binenbaum used EMR data to review the retinal examinations of 5,200 young children and concluded that the answer was, “No,” in a 2015 study published in JAMA Ophthalmology.

Another question is, “When do you give children eyeglasses?”

“It’s not so straightforward to know when to give glasses to young children, to know when the prescription is large enough that it could affect the child’s visual development if she doesn’t wear glasses,” Dr. Binenbaum said.

While guidelines based upon expert opinion exist, Dr. Binenbaum was interested in what ophthalmologists were actually doing. No study had used patient-level data to analyze individual clinical decisions – a fruitful opportunity for daily practice to inform research, and vice versa.

Dr. Binenbaum leveraged EPIC, CHOP’s EMR system, to uncover patterns in glasses prescribing by CHOP ophthalmologists. Through EPIC, Dr. Binenbaum explained, you can see a child’s age, whether refractive error (how near-sighted or far-sighted they are) was measured, and whether or not glasses were prescribed. He looked back over a four-year period, at about 19,000 refractions, and was able to generate tables describing when doctors would prescribe glasses for different refractive errors at varying ages. Some clear patterns and insights emerged, including that our ophthalmologists often disagree with the recommended guidelines for infants, while they tend to agree more as children get older.

So how can clinicians take advantage of these research methods? Dr. Binenbaum explained to the audience that they can begin to think more about how they enter information into the EMR. This sort of forethought can pay off later, when clinician-researchers look to conduct research or improve quality in their practice.

“The more people use numbers instead of written descriptions, the more they can specifically define variables or fields that can later be analyzed without doing a lot of extra work, the easier it will be to do very large studies very quickly,” Dr. Binenbaum said. “By thinking about what kinds of questions you may want to answer in the future, you can start to become more disciplined in the way that you record things. Do it in a consistent, analyzable way."

Share This