Test–retest reliability of the multifocal photopic negative response

To assess the test–retest reliability of the multifocal photopic negative response (mfPhNR) of normal human subjects.

Multifocal electroretinograms were recorded from one eye of 61 healthy adult subjects on two separate days using a Visual Evoked Response Imaging System software version 4.3 (EDI, San Mateo, California). The visual stimulus delivered on a 75-Hz monitor consisted of seven equal-sized hexagons each subtending 12° of visual angle. The m-step exponent was 9, and the m-sequence was slowed to include at least 30 blank frames after each flash. Only the first slice of the first-order kernel was analyzed. The mfPhNR amplitude was measured at a fixed time in the trough from baseline (BT) as well as at the same fixed time in the trough from the preceding b-wave peak (PT). Additionally, we also analyzed BT normalized either to PT (BT/PT) or to the b-wave amplitude (BT/b-wave). The relative reliability of test–retest differences for each test location was estimated by the Wilcoxon matched-pair signed-rank test and intraclass correlation coefficients (ICC). Absolute test–retest reliability was estimated by Bland–Altman analysis.

The test–retest amplitude differences for neither of the two measurement techniques were statistically significant as determined by Wilcoxon matched-pair signed-rank test. PT measurements showed greater ICC values than BT amplitude measurements for all test locations. For each measurement technique, the ICC value of the macular response was greater than that of the surrounding locations. The mean test–retest difference was close to zero for both techniques at each of the test locations, and while the coefficient of reliability (COR—1.96 times the standard deviation of the test–retest difference) was comparable for the two techniques at each test location when expressed in nanovolts, the %COR (COR normalized to the mean test and retest amplitudes) was superior for PT than BT measurements. The ICC and COR were comparable for the BT/PT and BT/b-wave ratios and were better than the ICC and COR for BT but worse than PT.

mfPhNR amplitude measured at a fixed time in the trough from the preceding b-wave peak (PT) shows greater test–retest reliability when compared to amplitude measurement from baseline (BT) or BT amplitude normalized to either the PT or b-wave amplitudes.