The Association of Sleep Apnea and Cancer in Veterans

Sebastian Jara | 2018

Advisor: Amanda Phipps

Research Area(s): Cancer Epidemiology



To test the association between pre-existing obstructive sleep apnea (OSA) and subsequent cancer in a large, long-term cohort of veteran patients.


This retrospective matched cohort study included all veteran patients diagnosed with OSA 1993-2013 by ICD-9 codes in any Veterans Affairs facility and probability matched veteran patients without OSA. Cancer diagnoses were identified by ICD-9 codes at least two years after index OSA diagnosis in patients with OSA and matched inception year in patients without OSA. Treatment by continuous positive airway pressure (CPAP) prescription was identified by CPT and ICD-9 codes. We tested the association between OSA (diagnosis and treatment) and subsequent cancer using multivariate Cox regression with time since OSA diagnosis as the time axis, adjusting for potential confounders.


The cohort included 1,236,130 patients (593,429 with OSA and 642,701 without OSA) with mean age 58+/-13 years, predominantly male (94%), a minority obese (30%), and median follow up 3.9 years (range 2.0-24.1). The proportion of patients diagnosed with cancer was higher in those with vs. without OSA (6.2% vs. 3.0%, P<0.001). After adjusting for age, sex, year of diagnosis, smoking, alcohol use, obesity, and comorbidity, the hazard of developing cancer was more than double in patients with vs. without OSA (HR 2.02; 95% CI [1.98-2.05]; P<0.001). Adjusting for the same confounders, the hazard of developing cancer was 8% lower in patients prescribed CPAP vs. untreated patients, among OSA patients with a sleep test preceding their diagnosis (HR = 0.92, 95% CI: 0.88-0.97).


Pre-existing OSA was strongly associated with subsequent cancer diagnosis in this veteran cohort, independent of known cancer risk factors. Additionally, the provision of CPAP may be associated with a reduction in cancer incidence. Further analyses are ongoing, including analysis of specific cancer types, other OSA treatment modalities, and cancer-related mortality.