American Academy of Ophthalmology Web Site:
Original URL:

Glaucoma: Calculating the Risk
By Miriam Karmel, Contributing Writer

The medical management of glaucoma is haunted by the question, “Which patients can be expected to progress from ocular hypertension to manifest glaucoma?” Looking at combinations of demonstrable risk factors may go a long way toward answering that question.

Thinner central corneal thickness. Increased vertical cup/disc ratio. Elevated IOP. Higher visual field pattern standard deviation index. Older age.

In 2002, the Ocular Hypertension Treatment Study (OHTS) identified these as key risk factors for progression from ocular hypertension to glaucoma, along with the finding that treatment works. It was useful information. And some of it, like the corneal thickness finding, was even surprising. But what did any of it mean for clinical practice? Could it be used to improve patient care?

As David S. Greenfield, MD, noted, “OHTS identified six factors that in the OHTS population proved to be important with regard to the ability to identify patients likely to convert to glaucoma over the course of the five years of study. But it didn’t provide any guidance when faced with a specific patient that you’re examining.” Dr. Greenfield is associate professor of ophthalmology at the Bascom Palmer Eye Institute.

Research applied to the real world. Now, that guidance is available in the form of a calculator based on the OHTS data. “It’s the first time in glaucoma, and probably the first time in ophthalmology, that a risk calculator has been applied such that one can translate the results of a clinical trial to an individual patient,” Dr. Greenfield said, speaking of a calculator developed by Robert N. Weinreb, MD, and Felipe A. Medeiros, MD, professor of ophthalmology and assistant professor of ophthalmology, respectively at the Hamilton Glaucoma Center, University of California, San Diego. Dr. Weinreb is also the director of the Center.

That calculator, called STAR (Scoring Tool for Assessing Risk), was unveiled  at the Academy’s 2005 Annual Meeting. Its proponents say it is ready for prime time, and will lead to more appropriate treatment, and should do for ophthalmology and glaucoma what similar calculators have done for cardiology and heart disease for decades. “It will allow treatment to be targeted to those most in need of it, as well as avoid the costs and risks of treatment for those who least need it,” Dr. Weinreb said.

The STAR harvested the OHTS data to assess an individual patient’s risk of progressing from ocular hypertension to glaucoma within five years. Physicians who use it should have a better sense of which patients are at highest risk and which are at lowest risk of progression, Dr. Weinreb said.

The Cardiovascular Model
Risk assessment in medicine is nothing new. For more than two decades, cardiologists have been doing it with data from the 50-year-old Framingham Heart Study, from which was developed a model to estimate risk based on a number of variables. As a result, internists routinely measure their patients’ risk for heart disease by factoring such variables as smoking, hypertension, elevated cholesterol, obesity, diabetes and physical inactivity.

Glaucoma, some researchers argue, fits the hypertension/ cholesterolemia model in that it is a chronic disease with major known risk factors. Like cardiovascular disease, it is associated with multiple risk factors, which in combination may increase the likelihood of progression to glaucoma.

From good guesses to dependable data. The first attempts at risk assessment in glaucoma date back to the ’70s, when researchers considered using cup/disc ratio and IOP to predict progression to glaucoma. But such predictions were not supported by data. OHTS changed everything. “Without the OHTS data this would have been very, very difficult to do,” Dr. Weinreb said, explaining that it is one of a few independent study populations able to provide the supporting information.

Drs. Weinreb and Medeiros applied the Framingham heart model to the OHTS results. Then they validated the model on a subset of patients at the Hamilton Glaucoma Center at the University of California, San Diego, who participated in the NEI-sponsored Diagnostic Innovations in Glaucoma Study (DIGS) and who matched the OHTS profile. They published those results last October in Archives of Ophthalmology

“Finally, we’re getting into risk assessment for these folks, where in the past we simply went on one variable, and that was almost always IOP. Now we’re incorporating several different variables that the OHTS study pointed out as significant,” said George A. Cioffi, MD, chief of ophthalmology at Devers Eye Institute, Portland, Ore. He and colleague Steven L. Mansberger, MD, MPH, developed their own risk calculator, also based on the OHTS data. That calculator, which has yet to be validated, but which considers the same six factors as the STAR, can be seen at

How It Works
The STAR is an easy-to-use slide rule with which physicians enter six risk factors—the five mentioned above, plus diabetes mellitus, which in OHTS appeared to have a protective effect. The directions are printed on the STAR, which is offered free as a professional service by Pfizer Ophthalmics sales representatives.

Do the math. Users of STAR are walked through the risk assessment of a sample patient with the following profile: 62 years old; pattern standard deviation = 1.60 decibels; cup/disc ratio = 0.5; IOP = 27 mmHg; central corneal thickness = 510 micrometers; no diabetes.

After positioning the numbers as guided, the slide rule reveals that the probability of that 62-year-old patient converting to glaucoma within five years is 31 to 40 percent. That number was derived through a mathematical formula borrowed from the Framingham investigators and expressed like this:

Risk estimate = 1-0.906exp(PI), where PI = (0.0223 x age) + (0.1044 x IOP) + (1.1157 x PSD) – (0.0150 x CCT) + (2.7763 x vertical C/D ratio) – (1.0498 x diabetes mellitus) + 1.6904.

So should that patient be treated? Yes, according to a global risk thresholds chart printed on the calculator. The risk thresholds chart, originally published in AJO in 2004,2 is based on the consensus of a group of glaucoma experts, including Drs. Weinreb and Cioffi. They suggested treatment for any individual with a 15 percent or higher risk of converting to glaucoma within five years. The panel recommended monitoring patients with a risk of less than 5 percent. For patients in the mid-range of 5 to 15 percent, they advised physicians to “consider and discuss benefits and risks of treatment with the patient.”

Enter the Caveats
Paul R. Lichter, MD, expressed concern that use of a calculator could lead to rote treatment, while acknowledging that this was not the creators’ intent. “I fear that once disseminated in our field, risk calculators will be a strong force against individualized patient care for this set of patients and will be used to gently coerce ophthalmologists to treat their patients based on the values in the risk calculator,” he said. Dr. Lichter is professor and chairman of ophthalmology and visual sciences at the University of Michigan, Ann Arbor, and director of the Kellogg Eye Center. He conceded that a risk calculator “can add value to an ophthalmologist’s decision-making process,” but he fears it could lead to overtreatment. “It runs the risk of being used like risk calculators for systemic hypertension and coronary artery disease—namely, if the patient falls in the ‘treat’ category, then that’s the standard of care.”

Indeed, when contrasted with cardiovascular risk factors, only one of the risk factors in the glaucoma equation is modifiable, and that is IOP. So the parallels with heart disease could be overdrawn.

Simply another tool. But Dr. Weinreb said the calculator is not intended as a guideline for care. In fact, he shies away from the word. “I don’t like ‘guidelines.’ That says there’s something you should follow,” he said. “Generally, I prefer to avoid recommendations that dictate patient care.” Rather, he hopes the calculator will be used as a tool to help clinicians manage their patients based on their own judgment and experience. “I would envision that every physician has their own threshold.” Some might set the high-risk threshold at 25 percent, he said, while others may follow the 15 percent suggestion of the calculator’s designers.

“I think those cutoffs are reasonable, but please realize that these cutoffs are arbitrary,” said Dr. Cioffi, whose calculator observes the identical treatment thresholds. The real value in creating the categories is in helping to separate the high-risk patients from the low-risk, he explained. “When I talk about the Devers risk calculator [to doctors and patients], I always talk about it as high-, medium- and low-risk.”

Dr. Mansberger, a codeveloper of the Devers calculator, said, “These just simplify complex results for the clinician.”

Yet other caveats? Glaucoma risk calculators are easy to use and can aid in treatment decisions. But the current models cannot be used on every patient with ocular hypertension. For one thing, the calculator is useful only for patients with untreated ocular hypertension, and not patients with normal discs, IOP and family history.

Drs. Weinreb and Medeiros warn that their STAR scoring system is not intended to evaluate risk in treated patients or to reassess risk after treatment is initiated. What’s more, since their calculations are based on patients in the OHTS study, the tool’s use is limited to those patients who match the OHTS patient profile. In general, you don’t want to use a calculator based on OHTS data on patients not defined by the OHTS.

“Make sure that your patients look like the patients in the model you’re using,” said Dr. Cioffi. “Don’t overgeneralize.”

The ins and outs of OHTS. The OHTS inclusion criteria were BCVA of 20/40 or better, spherical refraction within ± 5 D, cylinder correction within ± 3 D, open angles on gonioscopy, age 40 to 80, and ocular hypertension in both eyes.

OHTS excluded those patients with secondary causes of high IOP (e.g., pseudoexfoliation, pigment dispersion syndrome, iridocyclitis or trauma), other intraocular eye disease, history of refractive surgery or other diseases possibly affecting the visual field. Also excluded were patients with any evidence of diabetic retinopathy, people who had values outside the OHTS range (24 to 32 mmHg in one eye; 21 to 32 mmHg in the other) or elevated pressure in one eye but not the other.

Its Use in Practice
The calculators may simplify a decision to treat, but they still leave plenty of room for clinical judgment. “The risk calculator should be used as an adjunct to, and not as a substitute for, clinical experience and judgment, and each physician is likely to have his or her own threshold for treatment,” said Dr. Medeiros. Other factors, such as patient’s overall health status, life expectancy and commitment to treatment should also be weighed against potential adverse events and cost, he said.

One size does not fit all. Dr. Cioffi agrees. “Other things have to enter the equation when patients are at moderate risk. There’s some intermediate zone where you are liable to take a lot of other things into account,” he said. “When it’s in the gray zone, the patient is actively involved. You might take into consideration the ease of lowering pressure, the age of the patient, the patient’s tolerance for other medications, or other things. But when they’re at one of the two extremes, you’re going to probably push harder if they have a high likelihood, or push less if they have a low likelihood. These are just ballparks. Different physicians will be more or less aggressive with those numbers.”

Attention must be paid. No matter how physicians use the numbers, risk assessment could improve clinical practice. Dr. Weinreb noted that studies have shown that clinicians don’t routinely examine and document the appearance of the optic disc, but the risk calculator leaves no option. “You have to examine the optic disc,” he said. “You have to measure CCT to use this tool. This will bring to everyone’s attention the im-portance of examining the optic disc.” He added that it also highlights the importance of an accurate baseline evaluation.

Another benefit of the risk calculator is its potential as a teaching tool. “Patients sometimes like to hear that number,” Dr. Cioffi said. “I think a lot of patients are reassured that it’s not just their physician guessing.”

Kuldev Singh, MD, professor of ophthalmology and director of the glaucoma service at Stanford University, agrees. “Rather than just being able to tell patients, as we did in the past, ‘You have high pressure and are at greater risk than someone with normal or average eye pressures,’ we can actually give them numbers that are patient-specific, as opposed to average numbers of overall risk,” he said. “Now you can have specific risks for specific individuals based upon characteristics other than just IOP elevation.” He added that a patient who better understands his or her risk is better able to make a well-informed decision regarding treatment recommendations.

Future Options
The glaucoma risk calculator may be ready for clinical practice, but it’s just a start. “This is the first version of the risk calculator,” said Dr. Weinreb, who anticipates that new data will lead to refinements. Dr. Greenfield agreed, and predicted that an enhanced model might include data on optic disc hemorrhage, abnormal HRT parameters, retinal nerve fiber layer atrophy and SWAP abnormalities.

Next stop: fine-tuning. In the meantime, Dr. Weinreb is already tweaking the first edition, particularly with regard to diabetes. Diabetes was included in the calculation because the OHTS found it to be slightly protective against progression. But the DIGS study, against which his calculator was validated, did not find that effect. “It may be reasonable,” he said, given the skepticism surrounding the data, “to not use the diabetes factor with the calculator until its role can be clarified.”

Dr. Singh agreed that the diabetes finding was surprising, even when one takes into account that only diabetics without retinopathy were allowed into the OHTS. “In the future, if other studies find that diabetes is not protective, refinements will most certainly be made to existing calculators,” he said.

One, two, many calculators. Aside from refining existing calculators, Dr. Mansberger predicted calculators designed with more appropriate risk estimates for different patient populations, for example, a calculator for the Japanese population, which is much different than the U.S., he said. “You’ll be able to use a model with similar predictive variables, but adjusted by the mean risk in a different ethnic population.”

And Dr. Singh hopes that the interest in calculators will stimulate the development of validated calculators that assess the risk of progressing from early glaucoma to visual impairment. “We just don’t have adequate data for such a calculator. Hopefully the interest in calculators that has been developed from OHTS will provide a stimulus for future clinical trials that can provide the risk information needed for such additional calculators.”

In the meantime, the OHTS study group is preparing another risk calculator based on the OHTS data, which will be validated with data from the European Glaucoma Prevention Study. It should be available when the Academy meets in Las Vegas next November.

1 Arch Ophthalmol 2005;123:1351–1360.
2 Am J Ophthalmol 2004;138:458–467.

How the Risk Calculator Could Change “The Number Needed to Treat”
Pick a number from one to 20. The OHTS found that preemptive treatment works: It can prevent a person with ocular hypertension from progressing to glaucoma. But it is a blunt instrument for the job, since 20 patients must be treated just to catch one who would have progressed.

Risk calculation, which attempts to home in on the particular subset of patients most likely to progress, could reduce the number of people now needed to begin treatment in order to prevent one individual from converting to glaucoma. “The ‘number needed to treat’ falls precipitously,” said Dr. Weinreb, if you treat only patients who are at higher risk.

By any number necessary? The number per se is of less importance to others. “I think [risk assessment] will lead to more appropriate treatment, but I don’t know if it will be more or less,” Dr. Singh said.

Dr. Cioffi concurs. “I don’t think it’s important whether treatment increases or decreases,” he said. “I think it will make treatment more appropriate. It is good for patients because you’re more likely to treat the right patients.”

Or no number, necessarily? But not everyone agrees that treating patients with ocular hypertension is even necessary. Dr. Lichter, for example, questions the applicability of a heart model to glaucoma. “Risk calculators for disease got their start with diseases that, if left untreated, can lead to death,” he said. “But for a disorder such as ocular hypertension or the earliest stage of glaucoma, there are no data to know the outcome on vision-related quality of life of not treating or delaying treatment up to a certain point in the disease’s progression.”

He added that the risk calculator is not calculating the risk of going blind but of losing “a few ganglion cells in the retina from the earliest glaucoma damage.” That, he said, is “far different from the risk calculators for cardiovascular disease, which give one the risk of really bad outcomes.” Better data are needed, he said, to describe the long-term effects of observing vs. treating ocular hypertension, and what we really need is a calculator to assess the risk of loss of vision-related quality of life.

After all things are considered. Nevertheless, Dr. Lichter added, “All of the above notwithstanding, risk calculators for glaucoma can have value when used as part of a decision-making process to treat or not to treat an individual patient.”

In fact, a warning to that effect, not unlike the one printed on cigarette packs, appears on the STAR slide rule. “This tool,” it says, “is not intended to substitute for your clinical experience and judgment. Appropriate treatment will vary based on individual patient characteristics.”


George A. Cioffi, MD  Chief of ophthalmology at Devers Eye Institute, Portland, Ore. Financial interests: None.

David S. Greenfield, MD  Associate professor of ophthalmology at the Bascom Palmer Eye Institute. Financial interests: None.

Paul R. Lichter, MD  Professor and chairman of ophthalmology and visual sciences at the University of Michigan, Ann Arbor, and director of the Kellogg Eye Center. Financial interests: None.

Steven L. Mansberger, MD, MPH  Glaucoma specialist at Devers Eye Institute, Portland, Ore. Financial interests: None.

Felipe A. Medeiros, MD  Assistant professor of ophthalmology at the Hamilton Glaucoma Center University of California, San Diego. Financial interests: Research for the STAR was supported in part by a grant from Pfizer.

Kuldev Singh, MD  Professor of ophthalmology and director of the glaucoma service, Stanford University. Financial interests: None.

Robert N. Weinreb, MD  Professor of ophthalmology and director of the Hamilton Glaucoma Center, University of California, San Diego. Financial interests: Research for STAR was supported in part by a grant from Pfizer.