Epidemiological studies have established positive associations between OSA and cardiovascular diseases and interventional trials have documented at least partial alleviation of the pathophysiological manifestations using continuous positive airway pressure (CPAP). Contemporary investigations indicate that OSA prevalence is two to three times higher than in reference population without cardiovascular diseases. Yet, despite the rapidly expanding knowledge on this issue, a large number of cardiac Veterans with OSA remain undiagnosed. Progress in reaching these goals is hampered by the lack of a cost-effective tool to identify, diagnose, and assess improvement in the many affected patients. Because educational interventions and paper guideline dissemination have little effect, we have developed a simple, accurate, and a point-of-care, computer-based clinical decision support system (CDSS) not only to detect the presence of sleep apnea but also to predict its severity. The CDSS is based on deploying an artificial neural network (ANN) derived from anthropomorphic and clinical characteristics.
The specific aims of this pilot research project were: 1) to assess the validity of a handheld clinical decision-support system in detecting OSA in patients with ischemic heart disease against polysomnography; and 2) to compare the diagnostic accuracy of the CDSS versus the Berlin questionnaire
We recruited 143 consecutive Veterans from the Western New York Healthcare System who met the following inclusion criteria: presence of coronary artery disease documented by coronary angiography or percutaneous coronary angioplasty, a history of acute myocardial infarction, or coronary artery bypass grafts surgery. Veterans who were already diagnosed with sleep apnea or those who were receiving CPAP therapy, oral appliances or who had surgical interventions for sleep apnea, were not eligible for enrollment in the study. Patients with history of neurologic disease, systolic congestive heart failure, or documented ejection fraction by echocardiogram or MUGA scan of <45% were excluded from participation. Demographic and anthropomorphic data were collected on enrollment. Participants were asked to complete the Berlin Questionnaire and to answer all eight questions of CDSS-software. At the end of the interview, Veterans were scheduled for an in-laboratory polysomnogram (PSG).
Out of 143 participants who signed an informed consent, 57 did not return for a sleep study or withdrew consent prior to PSG. Eight six males (68.4 11.4 years) were included in the analysis. The prevalence of OSA (AHI 5/hr) was 74% with a median AHI of 11.5/hr (range 0-90). When compared to PSG, the CDSS and the Berlin questionnaire achieved a sensitivity of 98.4% (95% confidence interval [CI] 91.6-100) and 71.8% (95% CI 59.2-82.4) and a specificity of 86.4% (95% CI 65.1-97.1) and 31.8% (95% CI 13.9-54.9) at a threshold value of AHI 5 with a corresponding area under the curve (AUC) of 0.92 (95% CI 0.84-0.97) and 0.52 (95% CI 0.41-0.63); respectively. At a threshold value of AHI 10, the CDSS and the Berlin questionnaire had a sensitivity of 98% (95% CI 89.4-99.9) and 72% (95% CI 57.5-83.8) and a specificity of 91.7% (95% CI 77.5-98.2) and 30.6% (95% CI 16.3-48.1) with an AUC of 0.95 (95% CI 0.88-0.98) and 0.51 (95% CI 0.40-0.62); respectively. The CDSSS had a higher predictive accuracy in predicting OSA compared to the Berlin questionnaire for both thresholds of AHI 5 and AHI 10 (p<0.001).
CDSS-dependent screening of underlying sleep disordered breathing in Veterans with ischemic heart disease represents a novel management strategy for reducing the large morbidity and mortality burden of cardiovascular diseases. Once integrated into daily practice, the CDSS may be used to establish a priori the likelihood of OSA before considering the use of a polysomnographic diagnostic study and to prioritize patients needing polysomnography.
External Links for this Project
Grant Number: I01HX000466-01
- Laporta R, Anandam A, El-Solh AA. Screening for obstructive sleep apnea in veterans with ischemic heart disease using a computer-based clinical decision-support system. Clinical research in cardiology : official journal of the German Cardiac Society. 2012 Sep 1; 101(9):737-44. [view]