How sociodemographic and hearing related factors were associated with use of hearing aid in a population-based study: The HUNT Study

To the best of our knowledge, this is the first study which has assessed association between sociodemographic and hearing related variables - including measured HI and self-reported hearing loss - and use of HA in a large population-based sample of older persons. In all, less than one seventh (14 %) of the participants used HA, but about two third had a HI (based on 0.5,1,2,4 kHz) higher than 25 dB over both ears. Use of HA was associated with higher degree of HI, being bothered by hearing loss and education above ground level in both men and women. Men without a spouse were less likely to use HA compared to men with a spouse. Women without spouse were not more or less likely to use HA than women with spouse or men without a spouse.

The prevalence of HA-use among community-dwelling older adults was in the present study equally low as other population-based studies (8–13 %) [9, 10, 12]. Thus, few older adults use HA in everyday life, even if the prevalence of HI in older adults is quite high [1‐4], and HA may enable more efficient use of the person’s remaining hearing through an increase in speech perception and recognition of non-verbal sounds [5].

It is expected that the need to improve hearing with use of HA increases as the degree of HI worsens. Even so, two review reports indicate that the evidence of an association between higher degree of HI and increasing chance of using HA after an audiological HA fitting process is mixed [22, 24]. In population-based studies HI may be estimated audiometrically or with use of self-reported questionnaire information about hearing acuity in various situations. Applied on the whole population, the former is quite resource demanding, whereas the latter is less reliable [33, 34]. We identified two population-based studies with a relatively limited sample size (N < 3000) which assessed the association between degree of measured HI and either HA-use or a 5 year incidence of HA-use as outcome. These studies have reported mean HI (based on the 0.5, 1, 2, 4 kHz frequencies) higher than 25 dB bilaterally associated with use of HA [12, 27]. In the present study with a larger sample size we were able to study the dose-response relationship along the whole distribution of hearing loss and found a linear relationship for medium frequency and a monotonous relation up till 70 dB for low frequency hearing. Unlike previous studies we were also able to examine the importance of low, medium and high frequencies for use of HA independently. No independent effect was found of high-frequency hearing levels for HA-use. For both women and men a 10 dB increase in mid –frequency hearing increased the odds for HA-use more than twofold. In fully adjusted analysis of men, higher low-frequency HI from 30 up to 50 dB increased the odds for HA-use, but when the low-frequency threshold was 70 dB or higher the odds for HA-use was even lower than for the reference group with low-frequency threshold less than 20 dB. When hearing is gradually lost, low-frequency and mid-frequency hearing is usually kept longer intact than high-frequency hearing and the low - frequency hearing is retained the longest. Thus, the results probably reflect that HA-use starts when HI involves the medium frequencies band, rather than the high frequency band. When the low frequency HI is 70 dB or higher the effect of HA-use may be limited, especially if the medium and high frequencies are even poorer.

Furthermore, as expected and previously found in clinical audiological studies and population-based studies, experienced consequences of HI were positively associated with HA-use [12, 22, 27], i.e. older adults being bothered by hearing loss were much more likely to use HA compared to those not bothered by hearing loss, also after adjustment for sociodemographic conditions and HI over three frequency bands. In addition, HA-use was in the same analysis not associated with increasing age. This finding is in line with the majorities of previous studies [9, 22, 24], but not all [12, 27]. Moreover, those having vocational and general education were more likely to use HA compared to those with lower education. This may indicate an undesirable pro-educated use of HA. Vikum et al have accordingly shown that people with high socioeconomic status use specialist health services more frequently than people with low socioeconomic status in the same area [35]. It is left to future studies to decide whether education is associated with help seeking behavior, access to treatment, or HA-use after fitting, however. Degree of income, which is related to level of education, has been found to be positively associated with HA provision but of less importance when treatment and HA is funded by the authorities [24]. Also, financial factors like costs of batteries and repairs have been found of importance for no use of HA after HA were fitted [36].

This large scale population study made it possible to study interaction between independent variables for HA-use, and we found an interaction between gender and marital status. Men having a spouse were twice as likely to be HA-users compared to men without a spouse, but there were no difference in use of HA between women with and without a spouse. We did not study the participants’ communication demands or the role of significant others in the present study, i.e. the spouse of the participants or other close relatives. However, others have found the communication demands and the role of significant others to influence audiological help seeking and/or HA-use [24, 36]. Women regularly have a higher pitch when speaking compared to men. Thus, it may be harder or more demanding for a man with HI without HA to communicate with his wife, than for women to communicate with their husband without HA even if their HI were about the same.

The strengths of this study are mainly the large study sample (10,499 persons aged 65 years or more) and the objective measure of HI. The participants were aware of this study's specific hypotheses during data collection, which limits the risk of reporting bias of HA-use.

The limitations of the present study need to be addressed. Even if the effects of the various audiological variables were adjusted by sociodemographic variables, we cannot rule out that there are confounders which we have not adjusted adequately for. In addition to communication demands and role of significant others, which we did not have information about, personality and cosmetic demands may influence HA acquiring and later HA-use [22, 24, 37]. Secondly, due to the epidemiological nature of the study, otoscopy was not carried out prior to the audiometry for the full sample. Previous results from otoscopy in a subsample, showed that non-normal otoscopy findings in 13.3 % of the subsample, only accounted for a mean threshold shift of 0.0 dB to 0.3 dB in this subsample, implying that missing otoscopy data have hardly affected our results much [30]. Detailed information about previous clinical audiological testing and HA acquiring is missing, therefore it was impossible to assess the proportion of persons who have acquired HA but do not use it and also impossible to explore factors associated with acquiring HA in the present study [27]. Thirdly, we do not have information about whether the HA-users were part-time or fulltime users. Even so, earlier studies have shown that self-reported use of HA correlates highly with objective measures of use [38].

There is usually some uncertainty related to the generalizability of results from population studies like ours. The hearing examination represented only a small part of the total examination program in the HUNT 2 study, so there is little reason to suspect that hearing acuity has substantially influenced the recruitment to the study. A reduction of the study sample due to missing data on single questionnaire items may have affected the estimated occurrence of HA-use somewhat, but has hardly more than trivially affected the observed relationships between the predictors and the outcome. Nord-Trøndelag county is considered to be quite representative of Norway, so there is also little reason to doubt that these results are valid nationwide. The extent to which the results can be generalized internationally is more questionable. It seems reasonable to assume that the results are essentially valid at least for rather wealthy societies with refund arrangements for HA similar to the Norwegian. A more important limitation to the generalizability is the age of the data, especially because the HA technology has developed considerably since 1998. Statistics on number of acquired HA per year have more than doubled in Norway since the time of our data collection, from 39 778 in 1998 to 84 606 in 2015 (http://​www.​n-t-a-f.​org/​HA-statistikk.​htm). That does not necessarily show that the real HA use has been doubled, since we have a higher HI prevalence in the nation due to the ageing of the population, people may change HAs more frequently today (normally every sixth year) than twenty years ago and that it is more usual to choose HA for both ears now than 20 years ago, but it may reflect a higher proportion of HA users. But even if better technology may have caused a somewhat increased occurrence of HA-use during the last two decades, most of the reasons for using or not using HA have hardly changed fundamentally. Unfortunately, absence of more recent audiometrical data connected to use of HA makes it difficult to know the extent to which our results are valid for today’s hearing aid use.