Across 222,109 adults who underwent diagnostic sleep testing in Kaiser Permanente Southern California, apnea-hypopnea index (AHI) was tested against diagnoses, lab values, medications, procedures, healthcare use, behaviors, and symptoms. After full adjustment, 169 prevalent phecodes and 16 incident phecodes cleared Bonferroni significance. Positive associations clustered on hypertension, impaired glucose metabolism, hyperlipidemia, obesity, and heart failure, with matching laboratory, medication, and cardiac-measurement signals. Data-driven phenotypes combining physiology, symptoms, and comorbidities showed different disease incidence that AHI severity alone did not predict.
The cardiometabolic convergence confirms what sleep medicine has long suspected, and the value lies in its breadth: an unbiased scan that finds the same cluster from diagnoses, labs, and prescriptions is hard to dismiss as outcome-by-outcome cherry-picking. The more consequential signal is the mismatch between AHI and downstream disease. That fits growing criticism that AHI is a blunt severity metric and that hypoxic burden, arousal patterns, and symptom profiles may better predict who develops heart disease. Practically, adults with snoring, daytime sleepiness, or resistant hypertension have reason to ask about testing, and anyone diagnosed with OSA merits attention to blood pressure, glucose, and lipids. Limits matter, though. The cohort consists of people referred for sleep testing, so it skews toward symptomatic, higher-risk patients. Observational EHR data cannot separate OSA from shared drivers like obesity, and only 16 incident associations survived, so causal claims are off the table. This is a preprint that has not been peer reviewed, and its findings may shift. Best read as a confirmatory, hypothesis-generating resource rather than a practice-changing result.