Fahad R Butt, Jawad Muayad, Thanansayan Dhivagaran, Tasha Miller, Salem Abu Al-Burak, Fatima Abid, Wissam Nassrallah, Brendan Tao, Yusuf Ahmed, Clara C Chan
OSA with evidence of polysomnography was associated with incident KC after adjustment for measured confounders. The reverse direction showed a weaker, time-delayed association requiring cautious interpretation given detection bias. These findings support consideration of sleep evaluation in young adults with suspected corneal ectatic disease; they do not address whether treating OSA alters progression in established KC.
PURPOSE: To evaluate the bidirectional association between obstructive sleep apnea (OSA) and keratoconus (KC) in young adults using matched-cohort designs.
METHODS: Multicenter retrospective cohort study using the TriNetX Research Network. In the forward direction, adults aged 18 to 45 with OSA (ICD-10: G47.33) and evidence of polysomnography were compared with 3 propensity-score-matched comparators: polysomnography-evaluated controls, with and without age restriction, and a general encounter control. In the reverse direction, patients with KC and evidence of corneal imaging (H18.6 with topography, anterior segment OCT, or pachymetry) were compared with 3 parallel cohorts. Comparisons used 1:1 propensity-score matching (caliper 0.2) on demographics, body mass index strata, atopic/metabolic comorbidities, and health care utilization. Hazard ratios (HRs) were estimated at 5 and 10 years. Acute appendicitis served as the prespecified negative control outcome in all comparisons; ingrown nail was additionally evaluated in the general encounter comparator.
RESULTS: The forward primary analysis matched 48,954 patients per arm. The 5-year HR for incident KC was 2.01 (95% CI: 1.21-3.32; P = 0.006), stable at 10 years (HR: 2.03, 1.28-3.21), with concordant sensitivity HRs of 1.72 and 1.88. The reverse primary analysis matched 4350 patients per arm. The 5-year HR for incident OSA was nonsignificant (HR: 1.13, 0.91-1.40); the 10-year HR was 1.29 (1.07-1.55; P = 0.008). The negative control was null in all primary comparisons. The E-value for the forward primary HR was 3.45.
CONCLUSIONS: OSA with evidence of polysomnography was associated with incident KC after adjustment for measured confounders. The reverse direction showed a weaker, time-delayed association requiring cautious interpretation given detection bias. These findings support consideration of sleep evaluation in young adults with suspected corneal ectatic disease; they do not address whether treating OSA alters progression in established KC.