Principal findings
Only a limited number of studies are available providing prevalence estimates of hearing impairment for the general population in Germany. Moreover, the current epidemiological data are either associated with risk of bias or lacking generalizability. Overall, prevalence data on hearing impairment for the German population showed a wide variety ranging between 16 and 25% depending on the age, the study setting, the definition of hearing impairment and the method how hearing impairment was measured. For example, the reported hearing impairment was either based on (i) frequency-specific measurements by PTAs [
20], (ii) on a self-assessment (by telephone interviews) [
26] or (iii) retrospectively recorded using routine data (ICD coding) [
11].
Five primary studies claim that their data are representative for the entire population living in Germany [
11,
12,
20,
25,
26], but a critical methodological assessment did not (fully) prove these study conclusions. For example, one of these studies was conducted in West Germany in the years 1984 and 1985 [
25]. Since that time there have been major changes with regard to the national territory, the demographic composition of the inhabitants and the federal directives for the provision of medical aids. Thus, current representativeness of these study data is lacking. Another recent study reporting the age-dependent prevalence rate is limited to specific regions in Germany. Although this study extrapolated data to a national and European cohort to provide incidence rates up to the year 2025, methodological flaws and uncertainties resulting from any data extrapolation limit their validity [
20]. The study from the RKI used randomized samples from residents’ registration offices throughout Germany. However, the reported prevalence estimates of 18.8% (for minor hearing impairments) and 2.7% (for major hearing impairments) are based on telephone surveys which are associated with flawed data captures [
26]. Furthermore, people with known severe hearing impairment were excluded by the study author resulting in an underestimated estimate for major hearing difficulties. Another study carried out in general practitioners’ offices by Sohn and Jörgenshaus in the year 2001 utilized so-called quota sampling to achieve representativeness for Germany. Despite these approaches to control for prognostic factors, the study population consists of patients with a wide range of pre-existing conditions including ear disorders consulting a physician. Any transferability to all German inhabitants is therefore very questionable [
12]. Moreover, this study includes an unknown number of adolescents aged over 14 years of age. Considering that hearing disorders are less pronounced in younger than in older people, the reported overall prevalence rate may be underestimated. Furthermore, the study using ICD codes from a randomized sample of patient data to reflect diagnoses made by physicians reported an averaged prevalence rate (over all adult ages) of 6.6% [
11]. However, this estimate may be over- or underestimated because of two reasons: (i) beside the symptom hearing loss, ICD coding also includes the actual cause for this symptom and (ii) only patient data from those seeking medical advice because of subjectively perceived complaints were considered. Taking into account that patients suffering from hearing loss are often not aware of their impairment (or even deny it) and, therefore, do not consult a physician, the given prevalence rate may be underestimated [
29].
An estimate of the magnitude of these unrecorded cases may be provided by studies comparing prevalence data obtained by subjective self-assessments and making use of PTAs [
12,
19,
20,
25]. For example, HörMAT showed that approximately half of those over 60 years of age suffering from hearing loss were not aware of this impairment (41% (subjective self-assessment) versus 78% (PTA, ≥ 25 dB)) [
19]. In the study of von Gablenz et al. [
20] people over 80 years of age underestimated their hearing impairment (57% (subjective self-assessment) versus 72% (PTA, ≥ 25 dB)). The same observations were made in two other studies including much younger people (form the age of 14 or 15 up to 75 years of age or even more): 14% (subjective self-assessment) versus 19% (PTA, ≥ 40 dB) [
12] and 20.6% (subjective self-assessment) versus 26.8% (PTA, ≥ 30 dB) [
25]). Lacking information regarding the methodology of the subjective self-assessment excludes any quantification in terms of the severity of the perceived hearing loss.
We identified only two studies reporting on treatment coverage. However, none of these studies allowed us to judge the proportion of patients not wearing hearing aids although it would be indicated [
12,
20]. There is no literature available on treatment coverage by implantable hearing devices and cochlear implants in the general population. Furthermore, indications for pertinent operations are still a matter of current scientific discussion [
14].
Strengths and limitations of the systematic review
Overall, an epidemiological systematic review provides a useful approach to capture prevalence data on a certain clinical condition. However, some challenges in relation to the methodological assessment of the primary studies with respect to the risk of bias and the generalizability of the results exist. For example, there are no well-established tools to estimate the methodological quality of such studies. Therefore, based on published epidemiological studies we developed criteria to assess both risks of bias and generalizability for the included studies which may be of high value in future epidemiological research [
18]. Furthermore, the findings of our systematic review are based on a thorough and comprehensive literature search including hand searching for epidemiological studies on hearing loss in Germany. However, we are aware that our findings have several limitations related to the nature of our research work (systematic review): First, we could not identify sufficient study data to estimate the extent of hearing impairment for the general population living in Germany. Second, the risk of bias assessment revealed that the current prevalence data may be hampered due to inappropriate methods to capture such data (e.g., subjective self-assessments) and/or poor reporting making a thorough bias evaluation challenging. Generalizability of the results was limited due to hospital- or physician-based study settings based on patient data, regional or town specific participant selections and/or lacking up-to-datedness. Though the latest available prevalence data were captured between 2016 and 2017, they were limited to a special region in Germany and based on patient data [
19]. In another study, valid data were collected between the years 1984 and 1985, however, these data are the only representative for former West Germany [
25]. There would be even older prevalence data available, for example, data collected by Kessler and Hoffmann [
32], but we decided not to include such studies because of the very long time lapse between an associated demographical, socio-economical and technical changes since then.
Implications for clinical practice
This work has implications for researchers and those who use epidemiological data to help inform clinical and policy decisions. Overall, we revealed that (i) prevalence data on hearing loss are scarce and show methodological flaws. (ii) Prevalence data were broadly scattered reflecting different definitions of hearing loss, different methods of data capture, different not comparable age-groups and different settings; therefore, a comparison between studies was challenging. Further studies should be done by on the basis of a clear definition of hearing loss. This definition ought to be worked out in a first step. (iii) Moreover, available data are either not representative for the general in Germany living adult population (for example, most studies based their study sample on patient data or lacking currentness) or show methodological flaws limiting their validity. (iv) The rate of patients with hearing loss wearing hearing aids is very low (Fig.
2). Although this rate is increasing by age the total number of unaided patients is increasing as well (and more) with respect to the total number of hearing-impaired patients within elderly age classes. (v) We suggest a representative epidemiological study considering age-dependent frequency-specific definitions of hearing loss. This approach will also allow us to estimate the extent of the coverage of hearing aids. Particularly, in view of the high prevalence of the underlying diseases of hearing loss and the risks associated with untreated hearing impairment as well as the socio-economic costs such an epidemiological study is of great social importance.