Study design

The authors assessed 2,992 OCT scans from 474 glaucoma or glaucoma suspect eyes (241 patients). Factors influencing the reliability of RNFL thickness measurements (ΔRNFL, defined as the difference between the predicted and measured values) were assessed by statistical modeling techniques.

Outcomes

Reductions in signal strength correlated with reduced ΔRNFL. The effect on ΔRNFL was modest when the signal ranged between 10 and 3 (-0.67 to -1.25 ΔRNFL per 1-point decrease in strength) but was more dramatic when the signal was less than 3 (-15.7 to -16.34 ΔRNFL per 1-point decrease in strength). The decrease in signal strength was larger in eyes with severe glaucoma compared with mild/moderate disease.

Artifacts were also associated with a decrease in ΔRNFL, independent of signal strength. The impact of artifact was confined to the quadrant in which the artifact occurred.

Limitations

The measure of RNFL reliability is based on a predicted value. The 2-step modeling approach used here may have resulted in biased covariate estimates. Since the generalizability of the findings may be limited, further work assessing the effect of signal strength and artifact in a variety of patient populations and with different OCT machines will be needed.

Clinical significance

This study provides evidence-based guidelines on how to judge if an OCT scan shows the presence of worsening disease. A poor signal strength will yield a lower RNFL thickness, but may not truly be indicative of progression. Physicians should check each quadrant and discount measurements only from quadrants with artifacts.