More information needed to decide glaucoma status

Performing either structural or functional tests more frequently can provide more information about an eye’s glaucoma status than performing both types of test less frequently. Furthermore, performing both tests more frequently could show less discrepancy between the two types of test, but agreement between the two tests increases the redundancy of one of them, said Balwantray C Chauhan PhD, Department of Ophthalmology and Visual Sciences, Dalhousie University, Halifax, Nova Scotia, Canada. “The most important point is that if you’re not going to use a test for clinical decision-making, then don’t do it. Instead, do something that you’re comfortable with and do it more frequently and do a good quality exam,” he told the 12th European Glaucoma Society Congress in Prague, Czech Republic. Dr Chauhan noted that, with current diagnostic testing, changes in the optic nerve might become evident before visual field changes appear, or the opposite may occur. Numerous factors can play a role in the discrepancies. The type of testing to detect glaucoma most sensitively may also depend on the stage of the patient’s disease. One technique might be better at detecting early visual loss and another might be better at detecting later loss. “These measurements are a surrogate that are far away removed from what we really want to measure at the individual retinal ganglion cell level, at which point there should be a perfect correlation between the two,” he added.

The most important point is that if you’re not going to use a test for clinical decision-making, then don’t do it

The stringency of the criteria used can contribute to discrepancies between the two types of measurement. A study he and his associates conducted showed that when using intermediate criteria, glaucoma progression was detected in 28% of eyes with structural tests and in 27% of eyes with functional tests, but in only 10% of eyes with both structural and functional tests. When more conservative criteria were applied, only 14% had structural evidence of progression, only 14% had functional evidence, and only 3% had both structural and functional evidence of progression. Another contributor to the often poor agreement between structural and functional tests is signal processing errors, or noise, he said. He cited a study by Prof David Crabb, City University London, UK, who modelled structural change and functional change and simulated different degrees of noise. It showed that, in an eye where there was a good overlap when there was little noise, the simulated addition of noise progressively distorted the association between the results of the two types of testing. “This is depressingly like what we see in our data. I would like to be more optimistic and think that we are not measuring noise and independent indicators of disease progression,” Dr Chauhan said. TESTING FREQUENCY IMPROVES RELIABILITY Infrequent testing can also reduce the reliability of either type of test. A recently published US study showed that during the first two years after diagnosis of glaucoma, only 70% underwent perimetry, only 60% underwent testing with an imaging modality, and 20% underwent disc photography. “Quite staggeringly, around 10% of patients with a diagnosis of glaucoma have had no diagnostic testing done during those two years and that suggests that the diagnosis was made purely on the basis of intraocular pressure,” he pointed out. Dr Chauhan noted that one approach to integrating the findings from structural and functional testing is a Bayesian method, whereby one could use prior information from structural measurement as a prior probability, and convert that into a visual field map, using that, in turn, to model the rate of visual field loss. Studies using this approach show that the method improves results over using only one type of testing or the other, but also that the more frequently both types of testing are performed, the less of a difference there is between their findings. “This shows that, once you have a large amount of data, integrating this information may not be as valuable as having just one method of testing,” he added. Balwantray C Chauhan: bal@dal.ca