Visual Field Data Alone Inaccurately Estimates ICD-10 Glaucoma Severity Score

Using the location of visual field (VF) defects to provide International Classification of Disease, 10th revision (ICD-10) glaucoma scores frequently underestimates glaucoma severity, according to research published in the American Journal of Ophthalmology. Adding optical coherence tomography (OCT) information to the ICD-10 may improve diagnostic ability with respect to severity, according to the report.  

Researchers conducted a retrospective, cross sectional analysis consisting of 80 eyes from the Macular Damage in Early Glaucoma and Progression Study (mean age 68.2±8.6 years; 56.3% women; Clinicaltrials.gov identifier: NCT02547740). The study relied on 2 masked investigators to analyze OCT images, disc photos and visual field images. A second set of 2 graders classified glaucoma severity using visual field data and identified the VF regions on which they based their severity diagnosis (superior hemifield, inferior hemifield and central 5 degrees). A third pair of investigators evaluated glaucomatous damage using OCT to determine the presence of structural damage while blinded to VF findings and translated their findings to an equivalent ICD-10 classification based on the number of hemifields involved (0 hemifields, mild; 1 hemifield, moderate; 2 hemifields, advanced) or presence of central damage (advanced).

Among 80 eyes with glaucomatous optic neuropathy, VF identified 15 eyes with mild glaucoma, 23 with moderate glaucoma, and 42 with severe glaucoma using ICD-10 criteria. 

When OCT-based ICD-10 classifications were used to identify glaucoma severity, it categorized greater severity in 36% of cases. Analysis through OCT identified 5 eyes as having mild glaucoma severity, 7 with moderate glaucoma severity, and 68 with advanced severity. Agreement between OCT- and VF-based ICD-10 gradings was weak (κ=0.17±0.08; P =.006). Central damage detected by OCT was missed in 7 eyes classified as advanced by VF for a misidentification of central damage in 41% of the cohort. 

“The purpose of the current study was to test the hypothesis that based only on the 24-2 VF, this classification scheme will not accurately reflect the true extent of glaucomatous damage, which can be seen when structural (OCT) information is used,” according to the researchers. “We found that the VF-based ICD-10 severity score tends to underestimate the extent of glaucomatous damage to the optic nerve as determined by OCT.”

Study limitations include a retrospective nature and a high level of experience in performing VF tests under optimal testing conditions within the cohort.