Abstract
Observational studies have reported that hearing aid (HA) use is associated with a reduced risk of dementia diagnosis, suggesting a possible protective effect. However, extant observational studies do not explicitly model causal effects, while randomized controlled trials on the effect of HA on dementia exhibit short follow-up. Here we used self-report, hearing tests, and healthcare records in UK Biobank to design a hypothetical intervention for the effect of HA use on the risk of dementia diagnosis in people with incident hearing loss. HA users exhibited a higher risk of dementia diagnosis than nonusers (risk ratio: 1.43; 95% CI, 1.08-1.88). Associations between HA use and dementia diagnosis were robust across sensitivity analyses (risk ratio: 1.34-1.59), but adjustment for primary healthcare use (risk ratio: 0.77; 95% CI, 0.44-1.33) or primary and secondary care use (risk ratio: 0.68; 95% CI, 0.39-1.18) substantially decreased the observed effect. The decrease in effect estimates upon adjustment for primary (risk ratio: 1.30; 95% CI, 0.95-1.78), and primary and secondary healthcare use (1.30, 0.94-1.78) was smaller when participants with relatively early diagnoses of hearing loss were included in the sample compared to when they were not. While the findings are not conclusive, they suggest residual confounding by healthcare use and dating of hearing loss diagnosis in participants without primary care data in UK Biobank.