reports

A risk stratification model for adult patients with obstructive sleep apnea: development and evaluation

Toronto, ON, Canada: Institute for Clinical Evaluative Sciences.

Institute for Clinical Evaluative Sciences, Toronto, ON, Canada

STUDY HYPOTHESES: We hypothesize that the apnea-hypopneas index (AHI), currently used to establish severity of obstructive sleep apnea (OSA), is not by itself sufficient to accurately predict the study outcomes in individuals with OSA. We also hypothesize that an expanded set of factors including demographic, clinical patient characteristic and physiologic indexes are more accurate in predicting outcomes and will allow us to categorize adult OSA patients into groups at different risk (mild, moderate and severe) of experiencing negative consequences of OSA such as cardio-vascular events, death, diabetes, depression and cancer.

PRELIMINARY RESULTS: From 11,596 observations from the clinical dataset, 10,359 (62.2% men) were linked to administrative datasets. 21% had AHI<5; 27% 5-15 (mild OSA); 22% 15-30 (moderate OSA) and 30% over 30 (severe OSA).
OUTCOME:

• Time from the index sleep study to a composite cardiovascular (CV) outcome (myocardial infarction (MI), stroke, congestive heart failure (CHF), revascularization procedures, or death from any cause). Over a median follow-up of 70 months 1,191 (11.5%) participants had a composite CV event. Event-free survival at 10 years significantly decreased with AHI level (Figure 2).
• Time from the index sleep study to incident diabetes 8,843 subjects without diabetes at baseline identified through administrative dataset were included in analyses. Over a median follow-up of 68 months, 1,035 (11.7%) developed diabetes. Unadjusted for other factors, event-free survival at 10 years decreased with increasing AHI (HR=1.5; 95%CI: 1.4-1.6; p<0.0001).
• In a model, after adjustment for body mass index, age, smoking and income status, prior hypertension and number of comorbidities, the following OSA specific variables were significantly associated with incident diabetes in multivariable Cox regression model: increased AHI obstructive in REM sleep (standardized HR=1.3; 95%CI: 1.2-1.4), decreased total sleep time (TST) (standardized HR=1.12; 95%CI: 1.04-1.21) and self-reported daytime sleepiness (DS) (Yes/No) (HR=1.3; 95%CI: 1.1-1.4) (Figure 4). Model was well calibrated (Figure 5) and validated with Harrell’s C-index of 0.74.