Inter-device Variability of the Stratus Optical Coherence Tomography YANIV BARKANA, ZVIA BURGANSKY-ELIASH, YARIV GERBER, SHLOMO MELAMED, MEIRA NEUDORFER, ISAAC AVNI, ELISHA BARTOV, AND YAIR MORAD ● PURPOSE: To assess inter-device measurement variability with the Stratus optical coherence tomography (OCT). ● DESIGN: Evaluation of diagnostic test technology. ● METHODS: Eight eyes of eight healthy subjects were examined with four different Stratus machines in four medical centers during a period of five hours using fast retinal nerve fiber layer (RNFL) and fast macula proto- cols. Inter-device measurement variability and signal strength was assessed with standard deviation, coefficient of variance, and intraclass correlation coefficient (ICC). Spearman correlation coefficient was calculated between signal strength and thickness measurements. ● RESULTS: Statistically significant differences were not found for all macular parameters and all but one RNFL parameter. Mean signal strengths obtained with the four Stratus machines were significantly different; the newer the machine, the stronger signal strength it produced. Some RNFL parameters were moderately and statistically significantly correlated with signal strength. ● CONCLUSIONS: We did not observe statistically signif- icant differences in any macular and all but one RNFL parameter between the four Stratus OCT machines. Signal strength was significantly different between the machines, and significantly correlated with some RNFL parameters. Inter-machine variability in RNFL measure- ments may have clinical significance. Signal strength should be included in the assessment of measurements obtained on different machines and in serial examinations with a single Stratus machine during patient follow-up. (Am J Ophthalmol 2009;147:260 –266. © 2009 by Elsevier Inc. All rights reserved.) I MAGING OF THE TISSUES IN THE OCULAR FUNDUS WITH optical coherence tomography (OCT) has become widely used in the diagnosis and management of ocular disease. One prerequisite for meaningful interpretation of imaging results, both for initial disease diagnosis and for assessment of change over time, is low variability between measure- ments obtained by the same observer, by different observ- ers, and when repeated on different days. Some studies have suggested these components of measurement variance are low and satisfactory for routine measurements of macular thickness, retinal nerve fiber layer (RNFL) thick- ness, and optic nerve head parameters obtained with the time-domain version of OCT, the Stratus (Carl Zeiss Meditec, Dublin, California, USA). 1–3 Inter-device variability, the variability between different Stratus machines, is an additional very important factor. It should be known (and low) for proper interpretation of serial examinations of the same eye when performed at different institutions, when comparing a patient’s imaging result to a company-provided normative database that was obtained by different machines, and for the applicability and generalizability of studies performed worldwide. Ideally, all machines of the same model should produce exactly the same measurements when measuring the same eye by the same examiner. However, this may not be so, and inter-device variability may be a significant source of variabil- ity between studies from different clinical centers. Conceiv- ably, this factor can be an important reason why average values reported in the literature for healthy eyes have varied by 10% to 15%. 1– 8 It is rarely studied and reported, likely because of the difficulty of gathering a number of machines at a single location. In one recent study, significant differences were found between measurements of some parameters by two Stratus OCT instruments. 9 In another study, inter-device variability was evaluated for the scanning laser polarimeter. 10 The purpose of the present study was to assess Stratus OCT inter-device variability. We took advantage of the unique circumstance that in a relatively small area in the central part of Israel there are a number of OCT machines in close proximity. METHODS ● SUBJECTS: Eight eyes of eight healthy subjects were enrolled in this cross-sectional study. Volunteers were recruited among employees at the Assaf Harofe Medical Center. The examined eye was randomly selected in each subject. Subjects were enrolled in the study if they had no history or evidence of eye disease or surgery, 20/20 visual Accepted for publication Aug 11, 2008. From the Department of Ophthalmology (Y.B., I.A., Y.M.), Assaf Harofe Medical Center, Zerifin, Israel; Department of Ophthalmology (Z.B.-E., E.B.), Edith Wolfson Medical Center, Holon, Israel; Depart- ment of Ophthalmology (S.M.), Sheba Medical Center, Tel Hashomer, Israel; Department of Ophthalmology (M.N.), Tel Aviv Medical Center, Tel Aviv, Israel; and the Department of Epidemiology and Preventive Medicine (Y.G.), School of Public Health, Sackler Medical School, Tel Aviv University, Tel Aviv, Israel. Inquiries to Yaniv Barkana, Department of Ophthalmology, Assaf Harofe Medical Center, Zerifin 70300, Israel; e-mail: yaniv.barkana@ yahoo.com © 2009 BY ELSEVIER INC.ALL RIGHTS RESERVED. 260 0002-9394/09/$36.00 doi:10.1016/j.ajo.2008.08.008