Behavioral Tests for Audiological Diagnosis

This topic deals to a large extent with behavioral tests used for identifying the anatomical location

(“site”) of the abnormality (“lesion”) causing the patient’s problems—those traditionally referred to

as site-of-lesion tests. At this juncture it is wise to distinguish between medical and audiological diagnosis. Medical diagnosis involves determining the location of the abnormality and also its etiology, which involves the nature and cause of the pathology, and how it pertains to the patient’s health. In this sense, audiological tests contribute to medical diagnosis in at least two ways, depending on who sees the patient first. When the patient sees the audiologist first, they can act as screening tests to identify the possibility of conditions that indicate the need for referral to a physician. If the patient has been referred to the audiologist by a physician, then these tests provide information that assist in arriving at a medical diagnosis. In contrast, audiological diagnosis

deals with ascertaining the nature and scope of the patient’s auditory problems and their ramifications

for dealing with the world of sound in general and communication in particular. Diagnostic audiological assessment was traditionally viewed in terms of certain site-of-lesion tests specifically intended for this purpose. However, solving the diagnostic puzzle really starts with the initial interview and case history. After all, this is when we begin to compare the patient’s complaints

and behaviors with the characteristics of the various clinical entities that might cause them. Moreover, direct site-of-lesion assessment is already under way with the pure tone audiogram and routine speech audiometric tests. For example, we compare the pure tone air- and bone-conduction thresholds to determine whether the hearing loss is conductive, sensorineural, or mixed. This is the same question as asking whether the problem is located in the conductive mechanism (the outer and/or middle ear), the sensorineural mechanism (the cochlea or eighth

nerve), or a combination of the two. In addition, the acoustic immittance tests that are part of almost

every routine evaluation constitute a powerful audiological diagnostic battery in and of themselves.

In this chapter, we will briefly consider whether asymmetries between the right and left ears on the

pure tone audiogram help us to identify retrocochlear disorders; after which we will consider the

classical site-of-lesion tests. As discussed in  terms like acoustic neuroma or tumor, vestibular

schwannoma, and eighth nerve tumor will be used interchangeably. Following the traditional tests, we will cover the behavioral measures used to identify cochlear dead regions and some aspects of the assessment of central auditory processing disorders.

Asymmetries between the Ears

cochlear and eighth nerve lesions cannot be distinguished from one another based on the pure tone audiogram. In fact, any degree and configuration of hearing loss can be encountered in patients with retrocochlear pathology (e.g., Johnson 1977; Gimsing 2010; Suzuki, Hashimoto,

Kano, & Okitsu 2010). In spite of this, a significant difference between the ears is a very common finding in patients with acoustic tumors (e.g., Matthies & Samii 1997), so that the index of suspicion is raised when an asymmetry is found. Although we will be focusing here on asymmetries in the pure tone audiogram, it is important to stress that other asymmetries also raise the flag for ruling out retrocochlear pathology. Unilateral tinnitus (e.g., Sheppard, Milford & Anslow 1996; Dawes & Jeannon 1998; Obholzer, Rea, & Harcourt 2004; Gimsing 2010) and differences in speech recognition scores between the ears (e.g., Welling, Glasscock, Woods, & Jackson 1990; Ruckenstein, Cueva, Morrison, & Press 1996; Robinette, Bauch, Olsen & Cevette 2000) are among the other examples of asymmetries that become apparent early in the evaluation process.

However, many patients without retrocochlear pathology often have some degree of difference between their ears, as well; and there can be disagreement about whether a hearing loss is symmetrical or asymmetrical even among expert clinicians (Margolis & Saly 2008). It is therefore desirable to have criteria for identifying when a sensorineural hearing loss involves a clinically significant asymmetry. Thus, it is not surprising that various criteria have been suggested for identifying asymmetries in the pure tone audiogram that are clinically significant.

Many of these criteria are summarized in Table 1, although other kinds of approaches using

mathematical and statistical techniques have also been developed (e.g., Nouraei, Huys, Chatrath, Powles, & Harcourt 2007; Zapala et al 2012).

Some of the methods in Table 10.1 use an inter-ear difference of ≥ 15 dB as the criterion for

a clinically significant asymmetry, while others use require ≥ 20 dB. They also differ with respect to which frequencies are considered, how many of them are involved, and how the differences are calculated. The first two methods in the figure consider a difference between the ears to be clinically significant even if it occurs at just one frequency between 500 and 4000 Hz; and each of the next two considers only one specific frequency. The third set of methods require an asymmetry to be present for at least two frequencies in a range, but differ in terms of whether they can be any two frequencies or only ones that are adjacent to one another. Finally, the

last group of methods compare an average of the thresholds for a certain range of frequencies. Also notice that a few methods use different criteria based on gender, on whether the hearing loss is unilateral versus bilateral but asymmetrical, and even on whether the next step in the process would be magnetic resonance imaging (MRI) or auditory brainstem responses (ABR). So, with all these differences,

which criteria should be used?

At least part of the answer to this question is provided in the last two columns of the table, which

are based on the findings by Zapala et al (2012) for patients with hearing losses that were known to be either medically benign or associated with vestibular schwannomas. The sensitivity (also known as the hit rate) column shows the percentage of patients who had vestibular schwannomas who were correctly identified by the criteria. In contrast, the specificity column shows the percentage of patients who did not have tumors and who were correctly identified

by the criteria. For example, the criteria by Welling et al (1990) had 49.5% sensitivity and 89.4% specificity. This means that their method correctly identified 49.5% of the patients with tumors and 89.4% of those without tumors. This also means that 100 – 49.5 = 50.5% of the patients with tumors were missed by the method (false negatives), and that it also incorrectly flagged 100 – 89.4 = 10.6% of those who did not have tumors (false positives). Overall, we see that these provide up to ~ 50% specificity and roughly 14% specificity. Somewhat better performance can

be achieved with a statistical approach that allow the clinician to estimate the sensitivity and false positive rate for a given patient (Zapala et al 2012). This method uses the average difference between the ears at 250 to 4000 Hz (including 3000 Hz) and includes

adjustments for the patient’s age, gender, and noise exposure history.

Which method is “best”?

The answer to this question depends on how the comparison is made, and provides a convenient opportunity to introduce some useful concepts. One way to identify the optimal method is to

find the one that has the highest sensitivity and the highest specificity. In these terms, the overall topperforming criteria in the table were ≥ 15 dB for ≥ 2 frequencies between 250 and 8000 Hz (Ruckenstein et al 1996; Cueva 2004) with 49.9% sensitivity and 89.8% specificity, and ≥ 15 dB for 1 frequency between 500 and 4000 Hz (Welling et al 1990) with 48.5% sensitivity and 89.4% specificity.1 Another way to compare the methods is to graph the hit rate against the false alarm rate.2 (We will bypass the details, but interested students will want to know that this approach comes from the theory of signal detection, and the plot is called a receiver operating characteristic or ROC curve [see, e.g., Gelfand 2010].) With this approach, Zapala et al

(2012) found that their statistical method was the most effective method for distinguishing vestibular schwannomas from medically benign cases, followed by a close tie between ≥ 15 dB for the 250 to 8000 Hz average (Sheppard et al 1996) and ≥ 15 dB for the 500 to 3000 Hz average (Robinette et al 2000). The picture that emerges is that a significantly asymmetrical hearing loss provides up to ~ 50% sensitivity along with good specificity, and that the criteria just highlighted seem to be reasonable choices for use. Considering the limited sensitivity and high specificity of the criteria, a referral to rule out retrocochlear pathology is certainly warranted when a significant asymmetry is identified.

Reference :

• Gelfand, S. A. (2009). Essentials of Audiology. Thieme
• Roeser, R. J., Valente, M., & Hosford-Dunn, H. (2007). Audiology: Diagnosis. Thieme