DTTs in hearing screening

36

measurement error is inevitable. For example, Smits et al. (Smits, Kramer and Houtgast, 2006) estimated that in an elderly population, NHT

misplaced around 17% of the screened participants into the wrong hearing-status category. If self-report questionnaire data are compared with these hearing categories, only 83% correlation can be expected at best.

Objective measures capture only one aspect of hearing, whereas subjective evaluations are also affected by other factors such as

personality and everyday listening environments. Pronk et al. (Pronk, Deeg and Kramer, 2018) explored what contextual factors might explain the discrepancies between self-report measures and DTT results. They found that, for example, self-esteem, the number of chronic conditions, and the number of depressive symptoms may alter a person’s reference frame and influence the way they experience (objectively measured) hearing loss.

They also observed that gender and self-efficacy to initiate behavior seem to be related to coping strategies used compensate for the hearing loss.

37

The Flemish version of the DTT (Jansen et al., 2013) has been used as a school hearing screening test for elementary (5th grade) and secondary (3rd grade) school students in Flanders since 2016 (Denys et al., 2018).

Denys et al. (Denys et al., 2018) confirmed that for this age group, the test is feasible in a school setting and provides reliable SRTs. However, first grade elementary school children were too young to conduct the DTT efficiently and reliably as a self-test (Denys, Wouters and van Wieringen, 2021), even though the test, when conducted by a test administrator, was previously found feasible even for younger children (Koopmans, Goverts and Smits, 2018). Since the SRTs improve with age, age-specific referral criteria are needed for children up to the age of 12 years (Denys et al., 2018; Koopmans, Goverts and Smits, 2018).

The broadband versions of the DTT have been successfully used for screening for noise-induced HL (Jansen et al., 2013). In noise-exposed industrial workers, the correlation of hearing thresholds at 2–6 kHz with the DTT results was 0.86 for the Dutch broadband DTT version and 0.81 for the French broadband DTT version (Jansen et al., 2013). Jansen et al.

(Jansen et al., 2013) reported a sensitivity of 89% and specificity of 92% for detecting PTA >10 dB HL at 2, 3, 4, and 6 kHz. The bandwidth limited (0.3–

3.4 kHz) telephone version of the NHT was not as accurate in detecting high frequency HL, with a specificity of 94% and sensitivity of 55% for a group of NH and HI participants (Leensen, de Laat and Dreschler, 2011).

In addition to being sensitive and specific, a screening tool needs to be easily accessible to a large population. Remote DTT applications have provided results comparable to testing in the clinic (Vlaming et al., 2014;

Han et al., 2020). The current recommendation is to use headphones instead of a device´s loudspeakers when using DTT applications, as headphones allow for testing ear-specific SRTs and help to keep the

presentation level constant. In most studies, the quality of the headphones appears to have only negligible influence on the results, as simple in-ear headphones have been shown to provide results comparable to calibrated high quality headphones (i.e., Sennheiser HDA200 or TDH50-P) (Culling,

38

Zhao and Stephens, 2005; Buschermöhle et al., 2014; Potgieter et al., 2016).

Vlaming et al. (Vlaming et al., 2014) observed a small but statistically significant difference between good quality headphones and cheap

headphones or small loudspeakers for HI listeners but not for NH listeners.

The authors concluded that headphones should be recommended, since they likely reduce variability in the results.

Mobile applications improve access to hearing testing in areas where access to traditional hearing health care may be limited due to scarce resources or long and difficult distances. The South African hearZA is a good example of the applicability of mobile hearing screening in low or middle-income countries. After its initial launch, more than 30 000 people completed the hearZA over a period of 18 months (Potgieter et al., 2016;

De Sousa et al., 2018). The hearWHO application, released by the World Health Organization (WHO) in English in 2019, and in Spanish and Mandarin in 2021, has even wider uptake with more than 250 000 tests completed worldwide (De Sousa et al., 2022).

DTTs can also mostly circumvent the language barrier often associated with speech perception testing. Numbers are some of the first words learned in a foreign language, and listeners are constantly exposed to them in everyday life. Written numbers are also likely to be familiar to illiterate listeners, making them well suited for testing listeners from variable language and educational backgrounds. Language proficiency has been shown to affect speech perception in noise in NH and HI listeners (Warzybok, Brand, et al., 2015; Kaandorp et al., 2016; Scharenborg and van Os, 2019). For DTTs, the correlation with language proficiency has been significantly weaker (Warzybok, Brand, et al., 2015; Kaandorp et al., 2016;

Potgieter et al., 2018). Warzybok et al. (Warzybok, Brand, et al., 2015) observed smaller differences between native and non-native listeners with the DTT than with a sentence-level speech perception test, and concluded that listeners with at least intermediate language skills can perform the DTT reliably. Smits et al. (Smits et al., 2016) reported that young, NH, native Dutch-speaking listeners reach similar results in the US DIN as NH native

39

listeners. However, both studies evaluated only NH listeners, and noted that the results should be confirmed with HI listeners.

Large-scale data from HI listeners with variable language proficiency is available from South Africa (Potgieter et al., 2018) and from the UK Biobank (Taylor et al., 2020) dataset. For the hearZA, poor skills of South African English had a slight effect on the results, but no effect was observed for non-native listeners with at least intermediate language skills (Potgieter et al., 2018). The data from UK Biobank indicated language bias for the DTT, as the DTT results were more dependent on language ability and migration age than self-reported hearing measures (Taylor et al., 2020).

While the DTT was originally developed for hearing screening and to raise public awareness of hearing loss, it has proven well suited for collecting data on hearing in large cohort studies. Traditionally, objective hearing assessments have been difficult to include into large cohort studies, as threshold measurements require time, trained personnel, and calibrated equipment. DTTs enable cost-effective, large-scale objective data collection on hearing and have been included in multiple large studies in recent years.

One of the largest datasets to include a DTT is UK Biobank, which

contains SRTs for 160 995 participants (Moore et al., 2014). UK Biobank is a large-scale biomedical database and research resource that is globally accessible to researchers and contains in-depth genetic and health information from half a million UK participants. The dataset has provided unprecedented opportunities to assess, for example, the relation between speech-in-noise perception and cognition (Moore et al., 2014; Stevenson et al., 2021). Another large longitudinal study that has included DTT in its assessment battery is the Longitudinal Aging Study Amsterdam (LASA) (Huisman et al., 2011). The LASA data have been used to assess, for example, the decline of speech perception in the elderly over time, the correlation between changes in psychosocial health and speech perception in noise over time, and the relationship between cognitive decline and

40

decline in speech perception in noise (Nachtegaal et al., 2009; Pronk et al., 2019). The Netherlands Longitudinal Study on Hearing (NL-SH) is another large-scale study using DTT, and the researchers recently published a paper on DTT data over a 10-year follow-up showing that deterioration of speech perception in noise begins already in middle age and accelerates with age in patients over the age of 50 (Goderie et al., 2020).

2.2.4 DTTs in monitoring hearing rehabilitation results