Cumulative Atopic Disease Burden and Keratoconus: Cross-Sectional and Longitudinal Associations in the All of Us Research Program.

PURPOSE: To evaluate the association between cumulative atopic disease burden and keratoconus (KC) in a large, diverse United States cohort.

DESIGN

Longitudinal and cross-sectional cohort study.

PARTICIPANTS

A total of 283,040 keratoconus-free adults from the All of Us Research Program contributing 498 incident KC events over a median of 9.0 years (longitudinal cohort), and 508,527 adults including 741 with KC (cross-sectional cohort).

METHODS

A cumulative atopic burden score (0-3) was defined by the presence of asthma, atopic dermatitis, and allergic rhinitis. Cox proportional hazards and multivariable logistic regression estimated associations with incident and prevalent KC, adjusted for demographics, socioeconomic variables, and healthcare utilization, with a prespecified analysis restricted to participants aged 18-40 years at baseline and a matched case-control analysis in the age-appropriate window. Allergic conjunctivitis was evaluated as an independent covariate and, exploratorily, as a mediator.

MAIN OUTCOME MEASURES: Incident keratoconus (longitudinal) and prevalent keratoconus (cross-sectional).

RESULTS

Baseline atopic burden was associated with incident KC in a dose-dependent manner (per-condition trend HR 1.59; 95% CI, 1.30-1.94). Among participants aged 18-40 years at baseline, the association was stronger (trend HR 1.74; 95% CI, 1.31-2.32) and incident cases were diagnosed at a median age of 38 years. Cross-sectionally, the dose-response persisted after full adjustment (per-condition trend OR 1.33; 95% CI, 1.21-1.47). Adjustment for healthcare utilization attenuated the crude association, and a matched case-control analysis equalizing administrative presence further attenuated the gradient (OR 1.03; 95% CI, 0.77-1.38); this matched analysis was underpowered (105 cases) and did not exclude a residual effect. Allergic conjunctivitis was independently associated with KC (OR 2.81; 95% CI, 2.15-3.62). Atopy preceded KC coding in most co-affected patients.

CONCLUSIONS

Cumulative atopic burden was associated with incident keratoconus in a reproducible, dose-dependent manner, with atopy coding preceding keratoconus in a majority of co-affected patients. Differential healthcare ascertainment likely contributes to this association, although the matched analysis quantifying it was underpowered and did not exclude a true residual effect. These hypothesis-generating findings identify multi-condition atopy as a candidate marker warranting prospective, imaging-based evaluation.