Long-term subjective loneliness in adults after hearing loss treatment

Introduction

Loneliness among older adults is a highly prevalent public health problem, with 20 to 40% of older adults reporting feeling lonely at times (Luo et al. 2012; Savikko et al. 2005; Theeke 2005). Loneliness in older adults is associated with increased dementia (Wilson et al. 2007), blood pressure (Hawkley et al. 2010), and overall mortality (Penninx et al. 1997), and thus recognizing modifiable risk factors that may reduce loneliness in older adults if of serious concern. Previous studies have demonstrated hearing loss is independently associated with loneliness, which may be explained by reduced social engagement or impairment of communicative functioning (Sung et al. 2016). Previous research has demonstrated that hearing aids (HA) and cochlear implants (CI) may reduce loneliness symptoms in the short-term. Significant decline of self-perception of loneliness after 6 weeks of HA use has been documented (Weinstein et al. 2016). Likewise, there is significant reduction in 6-month measures of loneliness in CI users (Contrera et al. 2017). The largest improvement in loneliness symptoms after amplification appears to be observed in individuals with the greatest baseline loneliness and hearing impairment (Weinstein et al. 2016; Contrera et al. 2017). However, long-term (>1 year) outcomes on loneliness from treating hearing loss with HAs and CIs remain understudied and this is one of the first studies to present findings over an extended period of time. In this five-year follow-up to the Studying Multiple Outcomes After Aural Rehabilitative Treatment (SMART) study (Li et al. 2017), we assessed the long-term impact of HAs and CIs on loneliness in older adults. We hypothesized that long-term use of hearing technology is associated with reduced loneliness in older adults with hearing loss.

Methods

Study Participants

In this five-year follow-up study, we recruited participants from a cohort who completed the SMART study (Li et al. 2017). At baseline, 156 participants were enrolled from an eligible 564 patients who presented to the Johns Hopkins Listening Center for evaluation for a HA or CI between August, 2011 to January, 2014. Participants were eligible to participate in SMART if they met the following criteria: (a) ≥50 years of age, (b) English speaking, (c) receiving a HA for the first time or with previous minimal use (<1 hour/day) or a first CI, (d) diagnosed with post-lingual hearing loss, (e) aural-oral verbal communication as a primary communication modality, and (f) signed informed consent. The analytic cohort (n = 115) for this study consisted of individuals who had completed a baseline session, received treatment, and completed at least one follow up session.

Of the 156 participants at baseline, 22 participants did not provide written consent to be contacted for future follow-up. A review of Johns Hopkins electronic medical records revealed that four further SMART participants were known to be deceased and one never received treatment. Therefore, 129 original SMART study participants were contacted to participate in this follow-up study. Contact information of known non-deceased and previously consented participants were updated through review of Johns Hopkins electronic medical records as appropriate. One hundred eight participants were initially contacted by email and 22 participants were contacted by post mail based on their preferred mode of contact. Email reminders were sent at one week to those with an email address. At two weeks, we post mailed questionnaires to all participants who had responded and followed with telephone reminders at four weeks. We learned six of those contacted were deceased through family while reestablishing contact. Questionnaires were received between June to August, 2017 from 81 participants (62% response). Completion of questionnaires served as consent for this follow-up study.

Participants in this follow-up were offered a USD $20 check for participation. All study procedures were reviewed and approved by the Johns Hopkins Institutional Review Board (Baltimore, Maryland; Study Number: NA_00045877).

Loneliness Assessment

The UCLA Loneliness Index-Revised is a 20-item scale that assesses how often a participant feels disconnected from others (Russel 1996). The scale asks if the participant has felt certain emotions, such as ‘left out’ or ‘isolated from others,’ ‘never,’ ‘rarely,’ ‘sometimes,’ or ‘often.’ The index is scored from 0-60 with lower scores indicating less loneliness. This assessment was previously completed by participants in the original SMART study at baseline, 6-month, and 12-month time points and was completed approximately 5 years after baseline as part of this follow-up study.

Statistical Analysis

Data were collected and managed using REDCap electronic data capture tools hosted at Johns Hopkins University (Harris et al. 2009). All analyses were performed in R version 3.4.1 (R foundation for Statistical Computing, Vienna, Austria). Generalized estimating equations (GEEs) assuming an independent working correlation structure were used to model the population average outcome over time, accounting for the correlation between repeated measurements. While GEEs give consistent estimates of the population average even when the correlation structure of the outcomes is incorrectly specified, biased results can be obtained when data are missing and a complete case analysis is performed. To mitigate bias from non-ignorable missingness, missing covariates and outcomes were imputed using multivariate imputation with chained equations (MICE) using the MICE package (van Buuren et al. 2011). Predictive mean matching was used to impute missing values from a prediction model from the five observed values nearest to the predicted value. Convergence and fit were assessed after 20 iterations. GEE results were obtained and combined over 50 imputed datasets.

Results

Baseline demographic features of participants by treatment group are presented in Table 1. The analytic cohort consisted of 51 CI users and 64 HA users. This cohort is predominately non-Hispanic White (90.4%) and well educated (84.3% with a college or associate degree or higher). We found no significant differences in age, sex, race, history of noise exposure, and smoking, medical, and noise exposure history between treatment groups. The HA cohort is on average more educated and has a higher income than the CI group (p=0.012 and p=0.015 respectively). Figure 1 shows the mean UCLA scores at baseline, 6-month, 1-year, and 5-year time points. UCLA scores are similar to those at baseline for the HA group at all follow-up time points. UCLA scores are significantly reduced at 6-months (mean change: −3.76, 95% CI −5.80, −1.72) and 12-months (mean change: −2.99, 95% CI −5.73, −0.25) from baseline for the CI group, but return to near baseline at 5-years (mean change: −0.15, 95% CI −3.78, 3.48).

Table 1:

Summary baseline demographics of SMART study participants by treatment group (n=115)

Characteristic	Cochlear Implant (n=51)	Hearing Aid (n=64)	p-value
Age [interquartile range]	69 [64.2, 78.5]	71 [63.5, 75.8]	0.930
Sex
Female	21 (41.2%)	26 (40.6%)	1.000
Male	30 (58.8%)	38 (59.4%)
PTA* [interquartile range]	69.4 [64.4, 81.6]	36.2 [27.5, 42.8]	<0.001
Race
Non-Hispanic White	46 (90.2%)	58 (90.6%)	1.000
Other	5 (9.8%)	6 (9.4%)
Education Level
≤High School	12 (23.5%)	4 (6.2%)	0.012
College or Associate	22 (43.1%)	26 (40.6%)
>College	16 (31.4%)	33 (51.6%)
Unknown	1 (2%)	1 (1.6%)
Household Income
<$25,000	6 (11.8%)	2 (3.1%)	0.015
$25,000-$74,999	18 (35.3%)	9 (14.1%)
$75,000- $99,999	8 (15.7%)	9 (14.1%)
>$99,999	17 (33.3%)	34 (53.1%)
Smoking
Current or Former	30 (58.8%)	26 (40.6%)	0.087
Never	20 (39.2%)	36 (56.2%)
Unknown	1 (2%)	2 (3.1%)
Medical History
Hypertension	27 (52.9%)	35 (54.7%)	1.000
Diabetes Mellitus	7 (13.7%)	17 (26.6%)	0.163
Noise Exposure
Firearms	19 (37.3%)	23 (35.9%)	1.000
Occupational	19 (37.3%)	14 (21.9%)	0.094
Leisure	6 (11.8%)	9 (14.1%)	1.000

^*Pure tone average (PTA) indicates better-ear 4-frequency (0.5, 1, 2, 4kHz) threshold (dB HL).

Figure 1:

Mean University of California, Los Angeles (UCLA) scores of SMART Participants from at the 4 study time points (baseline, 6 months, 1 year, and 5 years) (N=115). Error bars depict 95% confidence intervals.

Discussion

In this follow-up study of older adults with age-related hearing loss treated with HAs or CIs, significant improvement in loneliness measurements were noted over one year in individuals with a CI. Loneliness measurements after five years were near those at baseline for both HA and CI recipients. Our results meaningfully demonstrate no significant increase in loneliness long term with hearing interventions, particularly as we may expect older adults to naturally be lonelier with time. It has been demonstrated that older adults get lonelier as time passes, likely due to death of aged peers, withdrawal from social activity, and increased functional limitations (Dykstra et al. 2005; Luhmann et al. 2016; Qualter et al. 2015). CI users, unlike HA users, demonstrated improvement in loneliness in the short term. The CI cohort also had greater baseline loneliness and hearing impairment than the HA group. As reduction in loneliness symptoms may result from improvements in speech understanding and quality of life (Contrera et al. 2017), individuals with poorer baseline hearing and social functioning may be those with the greatest improvement in loneliness.

Long-term outcomes of hearing loss treatment on loneliness remains understudied and results of prior studies are mixed. Contrary to the present study, Knutson et al. (1998) observed sustained reduction in perceived loneliness as measured by the UCLA Loneliness Scale 4.5 years after cochlear implantation in 37 older adults. While that study observed sustained reduction in perceived loneliness after cochlear implantation, 26% of the initial sample was excluded from their analysis due to missing data (Knutson et al. 1998). Bosdriesz et al. (2017) demonstrated that CI users with an average of 5.5 years of use were less lonely than their HA peers with average 15.2 years of HA use, though this analysis did not compare loneliness within the groups over time. These dissimilar findings may be attributable to differences in sample size and standardized measures of loneliness.

More proximal outcomes in hearing treatment, such as hearing handicap or communicative functioning, are significantly improved and sustained in long-term treatment of hearing loss with a cochlear implant (Helbig et al. 2016). It is likely that hearing loss treatment may more directly impact hearing handicap and communication capacity than it does a distal outcome such as loneliness as psychosocial factors, which were not considered in the current analysis, may influence subjective ratings of loneliness.

One limitation of this study is that it did not enroll an untreated cohort, and as such the contribution of normal aging on observed trends remains unknown. Even though loneliness in HA and CI users at five years was near that of baseline, we cannot know if participants would be lonelier without intervention. Our study may also not be fully generalizable as our cohort is primarily Caucasian, educated, and presenting to a clinic for hearing loss treatment. Likewise, the UCLA Loneliness Scale is designed to measure one’s subjective feelings of loneliness and is not devised to explain why that loneliness exists.

Missing data also posed a limitation with this study, particularly with a cohort of older patients with sensory loss participating at multiple, spaced time-points. Missing data were assumed to be missing at random and were imputed from observed data. This reduces the power of the study as a large portion of our data set across each time point was imputed. However, as noted, missing data are a key issue in other long-term studies of the impact of hearing loss treatment on loneliness.

This study found no long-term increase in loneliness in HA and CI users. Short-term reduction in loneliness in CI users was demonstrated. Our results contribute to the expanding medical literature examining treatment of hearing loss to mitigate loneliness in older adults. Future directions include more long-term studies on loneliness outcomes to explore the observed pattern further and randomized control studies to determine the causality of hearing loss treatment with trends in loneliness.

Acronyms:

CI

cochlear implant

FMI

fraction of missing information

GEE

generalized estimating equation

HA

hearing aid

MICE

multiple imputation with chained equations

PTA

pure tone average

SMART

Studying Multiple Outcomes after Aural Rehabilitative Treatment

UCLA

University of California, Los Angeles