Associations Between Pre-Fitting Factors and 2-Year Hearing Aid Use Persistence, Derived From Health Records and Post-Fitting Battery Order Data of 284,175 US Veterans.

Objectives: To examine associations between factors in domains of general health, hearing status, and demography, and subsequent long-term persistence of hearing aid (HA) use. By examining only non-modifiable factors available before HA fitting, we focus on potential indicators of a need for additional clinical effort to achieve satisfactory outcomes.

Design: The initial dataset consisted of Electronic Health Records spanning 2012-2017, for all (731,231) patients with HA orders from U.S. Department of Veterans Affairs audiology between April 1, 2012 and October 31, 2014. Applying inclusion criteria (valid HA use persistence data, complete audiograms, age ≥50 years, audiometric pure-tone average (PTA) ≥25 dB HL, 5-year clearance period for health conditions) and excluding records with codes for cochlear implants, the final sample was 284,175 patients. Independent variables encompassed audiological (PTA, PTA asymmetry, audiogram slope, audiogram complexity, new versus experienced HA user), health (dementia, mild cognitive impairment, other mental health conditions, multimorbidity, in-patient episodes), and demographic (age, race, ethnicity, partnership status, income, urban-rural home location) domains. The outcome measure was HA use persistence at 2 years post-fitting, based on battery orders within 18 months preceding the 2-year mark. Multiple logistic regression modeling was applied with HA use persistence at 2 years post-fitting as outcome. Continuous variables were discretized; missing data were imputed.

Results: After adjusting for covariates through the regression model, a significant positive association was found between PTA severity and HA use persistence, while PTA asymmetry, audiogram slope, and audiogram complexity were negatively associated with persistence. Being a new HA user, being diagnosed with dementia or other mental health conditions, and increased multimorbidity were all associated with reduced persistence. Persistence peaked at ages 70 to 79, and decreased for non-White races, Hispanic ethnicity, and those not married. No significant associations were found between persistence and tinnitus, urban-rural living, or mild cognitive impairment (when the model included dementia or other mental health conditions).

Conclusions: Pre-fitting personal factors other than audiological ones have independent, and summatively major, influence on HA use persistence. While not being modifiable, some are potentially usable as flags for a differentiated approach to patient management.