Reproducibility of the mfERG between instruments
Published Web Locationhttps://doi.org/10.1007/s10633-009-9171-z
Purpose First, to examine both the reproducibility of the multifocal electroretinogram (mfERG) recorded on different versions of the same instrument, and the repeatability of the mfERG recorded on a single instrument using two different amplifiers. Second, to demonstrate a means by which multicenter and longitudinal studies that use more than one recording instrument can compare and combine data effectively. Methods Three different amplifiers and two mfERG setups, one using VERIS™ 4.3 software (mfERG1) and another using VERIS™ Pro 5.2 software (mfERG2), were evaluated. A total of 73 subjects with normal vision were tested in three groups. Group 1 (n = 42) was recorded using two amplifiers in parallel on mfERG1. Group 2 (n = 52) was recorded on mfERG2 using a single amplifier. Group 3 was a subgroup of 21 subjects from groups 1 and 2 that were tested sequentially on both instruments. A fourth group of 26 subjects with diabetes were also recorded using the two parallel amplifiers on mfERG1. P1 implicit times and N1-P1 amplitudes of the 103 local first order mfERGs were measured, and the differences between the instruments and amplifiers were evaluated as raw scores and Z-scores based on normative data. Measurements of individual responses and measurements averaged over the 103 responses were analyzed. Results Simultaneous recordings made on mfERG1 with the two different amplifiers showed differences in implicit times but similar amplitudes. There was a mean implicit time difference of 2.5 ms between the amplifiers but conversion to Z-scores improved their agreement. Recordings made on different days with the two instruments produced similar but more variable results, with amplitudes differing between them more than implicit times. For local response implicit times, the 95% confidence interval of the difference between instruments was approximately ±1 Z-score (±0.9 ms) in either direction. For local response amplitude, it was approximately ±1.6 Z-scores (±0.3 μV). Conclusions Different amplifiers can yield quite different mfERG P1 implicit times, even with identical band-pass settings. However, the reproducibility of mfERG Z-scores across recording instrumentation is relatively high. Comparison of data across systems and laboratories, necessary for multicenter or longitudinal investigations, is facilitated if raw data are converted into Z-scores based on normative data.