Participants
The Trøndelag Health (HUNT) Study is a large general health-screening study for the entire adult population of Nord-Trøndelag County, Norway. It consists of four surveys conducted between 1984 and 2019 [10]. We used data from two hearing surveys: HUNT2 Hearing (1996–1998) and HUNT4 Hearing (2017–2019).
HUNT2 Hearing included 17 of the 24 municipalities in the county. The participation rate was 63%, and a total of 51,529 persons attended. Valid audiometry and data from a questionnaire that was distributed to all participants and returned at the site of the examination were available for 49,594 participants. HUNT4 Hearing was carried out in the six larger municipalities, representing approximately two-thirds of Nord-Trøndelag County. The participation rate was 43%, and a total of 28,388 persons attended. The hearing studies are described in detail elsewhere [11, 12]. After excluding persons with missing questionnaires or non-valid audiometry, the final cross-sectional samples comprised 49,594 and 26,606 participants in HUNT2 respectively HUNT4. The number of subjects participating in both HUNT2 and HUNT4 Hearing were 12,115.
Outcome measures
We analysed two outcome variables, use of hearing aids and hearing aid benefits. The use of hearing aids was measured by the question, “Do you use a hearing aid?” (yes/no) and was obtained next to a filter-question about self-reported hearing loss: “Do you have a hearing loss that you are aware of?” (HUNT2), “Do you believe you have impaired hearing?” (HUNT4). Participants with self-reported hearing loss who missed data on the following question on use of hearing aids, were treated as no use of hearing aid if their measured hearing was normal (n = 1734 in HUNT2 and 340 in HUNT4). Otherwise, participants with self-reported hearing loss and missing hearing aid data were excluded (n = 1044 in HUNT2 and 366 in HUNT4).
The self-reported benefit of hearing aid was obtained only in HUNT4. Only participants first reporting use of hearing aid were included. It was measured with the single item, “How much help do you have from your hearing aid?”, in four categories: no help, some help, great help and removes all problems. Out of 1,751 users only 19 had missing information on self-reported benefit.
Explanatory variables.
We investigated several explanatory variables: demography (age, education, and income), hearing-related factors (hearing threshold, tinnitus), risk factors for hearing loss (occupational and impulse noise, head injury, recurrent ear infection) and birth cohort (only for the outcome hearing aid use).
Pure-tone average hearing threshold (PTA4) was determined as the average hearing thresholds of 0·5, 1, 2 and 4 kHz in the better-hearing ear. The severity of hearing loss was defined using the criteria for classification by WHO (WHO Stevens et al. 2013; Wilson et al. 2017) in 15 dB intervals from good hearing (< 20 dB) in the better ear, to total impairment (≥ 95 dB) in the better ear with disabling hearing loss defined as PTA4 ≥ 35 dB. Self-reported hearing loss was measured by the questions: “Do you have a hearing loss that you are aware of?” (HUNT2), “Do you believe you have impaired hearing?” (HUNT4).
Childhood-onset hearing loss (hearing loss diagnosed by an ear-nose and throat specialist as sensorineural or related to chronic suppurative otitis media, recurrent ear infections or otosclerosis) with PTA4 < = 25 dB was obtained in a subsample born between 1940 and 1980 from the School Hearing Investigation in Nord-Trøndelag (SHINT), an audiometric screening of all schoolchildren attending regular schools in the County of Nord-Trøndelag from 1954 to 1986 [13].
Tinnitus was defined as tinnitus that is experienced daily or almost always, with periods lasting more than 5 minutes (HUNT4) or 10 minutes (HUNT2) and experienced as bothersome.
We obtained the following information from national registers: education (primary school, secondary school, university < 4 years, university > = 4 years), employment status (not employed, employed), occupation (white-blue collar), pensionable income standardized on age and cohort, marital status and having children. White-collar/blue-collar occupation was based on the Norwegian version of the International Standard Classification of Occupations, ISCO88, with one-digit level codes 1–5 categorized as white-collar and codes 0, and 6–9 as blue-collar workers.
From similar questions in HUNT2 and HUNT4, we obtained estimates of risk factors for hearing loss: occupational noise (regularly been exposed to loud noise at your present or previous work [no/less than 5 hours/week, >=5 hours/week]), impulse noise (no, maybe, yes), recurrent ear infections (no, maybe, yes), hospitalization for head injuries (no, maybe, yes) and being dizzy (no, maybe, yes). The maybe category was coded as no exposure. We treated missing values on any of these risk factors as no exposure, which accounted for < 5% in each variable.
Statistical analysis
Statistical tests were calculated in Stata version 17.0 with 95% confidence intervals. The alpha level was set at 0.05 for all analyses.
Prevalence of hearing aid use. We presented the prevalence of use of hearing aids as a function hearing loss in the two cohorts. To provide representative population estimates for adults over 19 years of age in Norway we accounted for the age and sex distribution of the Norwegian population in 1997 and 2018, by applying weights obtained from Statistics Norway [14]. To compare the prevalence of hearing aid use and prevalence of hearing loss (≥ 35 dB) across age and cohort, we applied logistic models including sex, age, cohort and interaction between cohort and age. Probabilities along with their corresponding 95% confidence intervals was predicted using the margins command in Stata.
Predictors of hearing aid use and benefit. We used logistic regression to examine predictors of hearing aid use (yes/no) in the pooled cross-sectional sample. All explanatory variables were included in the model. Hearing threshold, age, education, and income were treated as continuous variables. To reveal cohort-specific associations we investigated two-way interactions between cohort and hearing (hearing threshold and tinnitus), age and sex. We also tested for two-way interactions between hearing threshold, age, and sex to explore if the association between hearing loss and use or benefit varied with age and sex.
We used ordinal regression to assess predictors of hearing aid benefit (four ordered categories) in the HUNT4 sample. We performed the same analyses as for hearing aid use, except the analyses including birth cohort.
To estimate the frequency of hearing thresholds that best predicted the use and benefit of hearing aids, we applied the same multivariable regression models replacing with each 8 frequencies from 250 to 8000 Hz as independent variables instead of the PTA4 threshold. Finally, we investigated the effect of permanent childhood hearing loss in a subsample consisting of individuals born between 1940 and 1980.
To account for the dependence in the pooled data resulting from subjects participating in both surveys, we employed cluster-robust standard errors, utilizing the sandwich estimator.