Meta-Analysis of Distortion Product Otoacoustic Emission Retest Variability for Serial Monitoring of Cochlear Function in Adults. Issue 5 (September 2015)

• Record Type: Journal Article
• Authors: Reavis, Kelly M.; McMillan, Garnett P.; Dille, Marilyn F.; Konrad-Martin, Dawn
• Is Part Of: Ear and hearing. Volume 36:Issue 5(2015:Sep./Oct.)
• Publication Date: 2015-09
• Subjects: Distortion product otoacoustic emissions -- DPOAE -- Meta-analysis -- Reference limits -- Test–retest; Hearing disorders -- Periodicals; Audiology -- Periodicals; 612.85
• DOI: 10.1097/AUD.0000000000000176
• Languages: English
• ISSNs: 0196-0202

Abstract

Abstract : Objective: Distortion product otoacoustic emissions (DPOAEs) have long been heralded as a means to objectively monitor cochlear function and increasingly are becoming a key component in hearing surveillance programs for individuals at risk for ototoxic- and occupational noise-related hearing loss. Yet clinicians are unsure how to define clinically meaningful shifts in DPOAE level. In this study, a meta-analysis approach is used to synthesize the DPOAE level test–retest literature to construct a set of DPOAE level shift reference limits that can be used clinically to define a statistically significant emission change. Design: The authors reviewed all published articles identified through a Medline search using the terms "Otoacoustic Emission Variability, " "Otoacoustic Emission Reliability, " "Otoacoustic Emission Repeatability, " and "Otoacoustic Emission Test Retest" restricted to DPOAEs, adults, and English language. Articles with DPOAE level data elicited by moderate stimulus levels for f 2 frequencies of 1000, 2000, 4000, or 6000 Hz were selected because these stimulus parameters were relatively well represented in the literature. The authors only included articles that reported the standard error of the measurement (SEM) or from which the SEM could be calculated. Meta-analysis was used to estimate the population mean SEM over the included studies. Models were fit separately for each f 2 primary and included days since baseline and study-specific random effects. Results: Ten DPOAE test–retest studies met inclusion criteria for this meta-analysis. The SEM values varied widely across published studies (0.57 to 3.9 dB) and were provided for relatively short time intervals (less than 15 days on average). Time, or days since baseline, was statistically significant at higher f 2 frequencies (4000 and 6000 Hz). From the model results, 90% reference limits specific to the f 2 and elapsed time between baseline and follow-up measurements were established. Reference limits provided correspond to negative (emission decrement) and positive (emission enhancement) shifts indicative of the amount of measurement variability that, using this approach, must be tolerated as "normal" fluctuations over time. Changes larger than the reference limits are considered significant and warrant follow-up testing. Conclusions: The meta-analysis presented provides reference limits that are appropriate for a set of specific f 2 frequencies and time intervals. The meta-analysis concerns the SEM statistic directly, so that any preferred reference limit can be computed from the results and should be predicated upon the screening application. The presumed advantage of this meta-analytic approach is increased precision relative to limits suggested by any of the individual studies included in the analysis. Abstract : The principal obstacle to utilization of DPOAEs for serial monitoring of cochlear function is the lack of published reference limits from which to compare patient results. This study utilized meta-analytic techniques to synthesize published DPOAE level test-retest data (reported as a standard error of the measurement) into frequency and time-specific reference limits to identify both decrements and increments in emissions. DPOAE shift reference limits are provided, which provide a 10% false referral rate. Other false referral rates also can be determined using the data provided. Caveats and limitations for use are discussed.