January - March 2021: The GREAT Accuscreen saga !!!!

  • Print

The Otoacoustic Emissions Portal is on-line for almost 21 years (we started in June 2000). In this period of time I have served as web editor and have never taken a strong position on a technology, paper, presentation etc. But this time I need to tell an interesting  story to the OAE community.

Our story begins in 2015-2016 when my lab undertook a project on neonatal screening with the Accuscreen device (at the time with Zoth industries, now Natus). After an initial beta-testing of a couple of hundrend infants and toddlers we observed by accident  that some subjects (7-8 y) presenting moderate sensorineural hearing losses when tested with AccuScreen they resulted as PASS. Since the OAE testing was not necessary for these patients, we have considered this as anecdotal evidence. The same year a few cases like these emerged, but they all considered as not as important.

Considering the diffusion of the Accuscreen device in the European EHDI (Early Hearing Detection and Intervention) programs I was very eager to validate the anecdotal evidence I have encountered in my Ferrara University clinic. So in the context of my collaboration with the Institute of Physiology and Pathology and of Hearing (Warsaw) and Dr. Wiktor Jedrzejczak   we designed a project to challenge various hypothesis of the data I had observed .

The objective of this study is to compare the data acquired from the new touchscreen Accuscreen device with the outcomes from the previous Accuscreen model, using the ILO-292 equipment as the gold standard. To identify the factors that might be contributing to different performances between the tested devices, very high-quality OAE data are required. To minimize issues related to OAE recording variability, which are often observed in neonates, the data for this study are recorded from adult subjects. This provided important advantages: (i) The software and hardware for the OAE data collection of all these devices is the same for neonates, children, and adults, and so the conclusions from an adult group can be easily extended to another population of interest; (ii) the adult OAE responses present significantly lower levels of noise, and so effects related to the recording device are more easily identified; (iii) for every subject it is possible to have supporting ata on hearing thresholds, i.e., data that are not easily available for neonates or young children. 2. Materials and Methods 2.1. Subjects Data were recorded from 94 adult subjects (51 women and 43 men) with ages ranging from 21 to 85 years (M = 48.7; SD = 14.5). Data were obtained from a total of 185 ears (101 female and 84 male). The group consisted of 61 ears with normal hearing and 124 ears with various degrees of hearing impairment ranging from mild to severe. More detailed information on the hearing thresholds is n a later section, where hearing thresholds are compared with the OAE screening results. Research procedures were approved by the Ethics Committee of the Institute of Physiology and Pathology of Hearing in Poland (IFPS/KB/06/2012), and all participants gave written informed consent. 2.2. Measurements The audiological testing included an otoscopic assessment, pure tone audiometry (PTA), tympanometry measurements, and OAE recordings. Hearing thresholds be. tter than 25 dB HL between 0.125–8 kHz were taken to represent normal hearing. All subjects presented normal middle ear function, as assessed by otoscopic examination and tympanometry (tympanometric peak pressures were between –100 and +100 daPa and peak compensated static acoustic admittance values were approximately 0.2–1.0 mmhos). TEOAEs were measured by three systems: the ILO-292 (software version 5.61, Otodynamics Ltd., UK) and two versions of the Madsen Accuscreen (Otometrics, Natus Medical Denmark ApS). All OAE testing was conducted inside an audiometric booth, and the testing order was randomized. The ILO-292 was selected as the gold standard for two reasons: (i) Results from this system have been reported in the majority of papers in the literature; (ii) the system has been tested extensively by the authors and has proved to provide good separation of normal and impaired ears [22–25]. Signal-to-noise ratios (SNRs) were collected using the standard nonlinear protocol (stimulus of 80 dB peSPL with a 50 Hz click rate and a 20 ms recording window), providing information at frequencies of 1.0, 1.4, 2.0, 2.8, and 4.0 kHz. The recording was qualified as a pass when the SNR in 3 of 5 frequency bands exceeded 3 dB [26]. In the case of Accuscreen, two versions were used. The older device, Accuscreen Pro (firmware: 1.07E1M, GN Otometrics, Denmark) ), which was produced between 2005 to 2010, will be referred to as Accu 1. The newer device, the Accuscreen OAE and ABR Screener (touchscreen, Type 1077, firmware 1.12.04877 SEU, GN Otometrics A/S, Denmark), which has been produced since 2010, will be referred to as Accu 2. The

The objective of this study was to compare the data acquired from the new touchscreen Accuscreen device with the outcomes from the previous Accuscreen model (the one with no touch screen) using the ILO-292 equipment as the gold standard. To identify the factors that might be contributing to different performances between the tested devices, very high-quality OAE data are required. To minimize issues related to OAE recording variability, which are often observed in neonates, the data for this study are recorded from adult subjects. This provided important advantages: (i) The software and hardware for the OAE data collection of all these devices is the same for neonates, children, and adults, and so the conclusions from an adult group can be easily extended to another population of interest; (ii) the adult OAE responses present significantly lower levels of noise, and so effects related to the recording device are more easily identified; (iii) for every subject it is possible to have supporting audiometric data on hearing thresholds, i.e., data that are not easily available for neonates or young children.

The analyzed data suggested that the first version of the Accuscreen device (not the touch-scrren models) generated responses very different than the responses from the ILO-292 device. Problems were correlated with hearing thresholds in the middle and high frequencies. This implies that subjects with moderate losses could pass as Normal hearing.  Analytically :

 Our data show that there were significant differences in TEOAE evaluation as assessed by the old and new versions of the Accuscreen device, even though they are from the same manufacturer. Differences were also observed when the Accuscreen devices were compared with the ILO-292 system. Agreement between both Accuscreen devices was gauged as moderate (77.3%, κ = 0.54), similar to the level of agreement (70.8%, κ = 0.44) between the older device (Accu 1) and the ILO. Only agreement between the newer device (Accu 2) and the ILO was gauged as acceptable (80.5%, κ = 0.61). The best point of comparison for our results that we could find was a Food and Drug Administration report that compared the new Accuscreen evaluated here (Accu 2) and the EchoScreen, which is the predecessor of the older version of Accuscreen (Accu 1). The disadvantage of this study was that it involved only a small study group. Nevertheless, for adult subjects, agreement of around 86% was achieved, which is slightly better than the agreements achieved here of around 70–80%.

In terms of audiometric results, the disagreement between the Accuscreen devices occurred most frequently in ears with higher thresholds in the 1–8 kHz range and where thresholds below 1 kHz were still within the norm. The newer version of the device  performed better than the old (Accu 1). We can only suppose that the TEOAE analysis method was improved in the new version. The biggest weakness of both devices is that they provided a pass to ears with a threshold elevation >25 dB at 8 kHz. However, this is not surprising in that the majority of TEOAE probes have a frequency response limited to 4–5 kHz. In addition, from previous studies, OAEs (especially TEOAEs) are generally known to perform best over only the 1–4 kHz range [31]. Indeed, the discriminant analysis revealed that the most important audiometric frequencies in terms of pass/refer results were 1, 2, and 4 kHz, which covers the TEOAE range. This range also includes the greatest slope for the group of ears for which the results were different .

When the SNRs from the ILO were used in the discriminant analysis, it showed that the most important frequencies were 1.4 and 2 kHz. This is probably related to the fact that adult TEOAEs present higher amplitude levels in the 1–2 kHz range. It is also relevant that these frequencies are also dominant in speech, and so the presence of OAEs in this range may indicate good speech understanding; however, this cannot be generalized to cases of auditory neuropathy or disturbances to auditory brainstem responses (ABRs). For ears with weak TEOAEs, a response above 2 kHz may not be present, and in the 1.4–2 kHz range it can be very small. This weakness may give rise to disagreement between different systems when different classification criteria are used (e.g., if the criterion is that 3 of 5, or 2 of 5, frequency bands have to reach a certain SNR). Generally, it can be said that problems with incorrect pass/refer classifications come from ears with downward sloping audiograms.

 You can access the whole paper here  and in this alternative MDPI link

  We were very convinced that the paper would make it into the mainstream Audiology journals, because studies on the efficacy of various OAE systems are very few in the literature. We have encountered an incredible resistance into publishing the data and the team which authored the paper has a vast experience in authoring and OAEs. I was particularly impressed with comments from various reviewers calling us to consider the well-known differences in ear-acoustics between infants and adults, which could explain our data.

The truth is that there are considerable differences between infants and adults not only in terms of acoustics but also in terms of their corresponding OAE responses. BUT the denominator is that the same system is advertised as being able to conduct OAE recordings in both populations. So if the same system can be used to access data in both infants and adults the data from a well controlled population reflect the performance of the recording device and they are NOT coloured by any differences in the acoustics between the two populations.