Perceived Hearing Symptoms Organize the Ear-Disease Comorbidity Network but Are Not a Causal Lever for Brain Health: A Triangulated Analysis

Objectives

To determine whether objective audiometric hearing loss, subjective hearing difficulty, and tinnitus occupy distinct roles in the population comorbidity network (perceived hub versus audiometric sparse connection), and whether the perceptually active layer exerts an independent causal effect on brain outcomes. Design, Setting, and Participants. Cross-sectional network medicine analysis of 18,939 participants from two nationally representative NHANES cycles (2011-2012, 2015-2016), triangulated with bidirectional Mendelian randomization and external longitudinal validation in CHARLS (China; N = 17,101). Results. Subjective hearing difficulty-not audiometric threshold-acted as the hub linking 16 disease nodes (3 ear, 13 chronic). Bidirectional Mendelian randomization found no robust genetically causal ear-brain association: the null for perceived hearing drew on 22-23 well-identified instruments, whereas objective audiometric loss was indexed by only 3 instruments and was therefore non-informative rather than affirmatively null; across 24 estimable pairs none survived Bonferroni correction (alpha approximately .002). In CHARLS, perceived hearing difficulty predicted incident self-reported memory problems. Conclusions. Perceived hearing symptoms, not audiometric thresholds, organize the ear-disease comorbidity network, yet this perceptually active layer showed no independent causal effect on brain pathology. Subjective ear symptoms are observational flags of multimorbidity-not causal levers-and should prompt comorbidity screening rather than be interpreted as modifiable causal targets.

Objective

To test whether objective audiometric hearing loss, subjective difficulty, and tinnitus occupy asymmetric roles in the population comorbidity network (perceived hub vs audiometric sparse connection), and whether the network-active (perceived) layer exerts an independent causal effect on brain outcomes.

Design, Setting, and Participants

Cross-sectional network medicine of 18,939 participants in two nationally representative NHANES cycles (2011–2012, 2015–2016), paired with bidirectional MR and external longitudinal validation in CHARLS (China; N = 17,101).

Exposures

Sixteen binary disease nodes (3 ear indicators; 13 chronic conditions). MR exposures: objective hearing loss, subjective difficulty, tinnitus.

Main Outcomes and Measures

Node degree centrality (network role); bidirectional MR of the ear–brain axis and the reverse direction; incident self-reported memory problems (CHARLS Cox).

Results

Bidirectional MR found no well-supported genetically causal ear–brain axis: for perceived hearing the null drew on 22–23 well-identified instruments, whereas objective audiometric loss was represented by only 3 instruments and was therefore non-informative rather than affirmatively null; across 24 estimable pairs, none survived Bonferroni correction (α ≈ .002);

Conclusions and Relevance

Perceived hearing symptoms, not audiometric thresholds, organize the ear-disease comorbidity network; yet this network-active layer exerts no independent causal effect on brain pathology. Perceived ear symptoms are flags of multimorbidity, not levers on brain health — arguing for symptom-based, not audiogram-based, screening.

Key Points

Question Do objective audiometric hearing loss, subjective difficulty, and tinnitus organize the comorbidity network differently, and is there a genetically causal ear–brain axis?

Findings In a cross-sectional network of 18,939 NHANES participants, perceived hearing difficulty—not audiometric thresholds—was the hub linking 16 disease nodes; bidirectional Mendelian randomization (24 ear–brain pairs) showed no well-supported genetically causal association.

Meaning Subjective hearing symptoms are observational flags, not causal levers, of brain health; they should trigger comorbidity screening rather than be read as modifiable causation.