ABSTRACT
The emergence of obstructive sleep apnea (OSA) as a common health condition associated with substantial expenditures of health care dollars has been accompanied by skepticism over whether this condition represents a “real disease” (1). Although there is irrefutable evidence that OSA contributes to excessive daytime somnolence, there is greater uncertainty over whether OSA contributes to the incidence or progression of the two health conditions of established public health importance, hypertension and cardiovascular disease (CVD). There are several challenges in establishing and evaluating the putative causal associations between OSA and health outcomes, which have fueled criticism of the current evidence base. Inferring causality between an “exposure” such as OSA and outcomes such as hypertension/CVD requires multiple lines of evidence. Data from observational studies, such as case-control or cross-sectional or longitudinal cohort studies, may be useful in quantifying associations and identifying subgroups in which the disease may be especially important. The validity of such inferences relates to the precision of the statistical estimates, which is a function of the sample size, the appropriateness of the study design and analytical methods, including the validity and reliability of the measurements of exposure and outcomes, and the extent to which findings are unexplained by selection, measurement, confounding, or other biases. There is greater confidence in such associations if they are consistent across studies and different populations, if consistent patterns are observed between increasing levels of exposure (severity of OSA) with severity or frequency of adverse outcomes (“doseresponse” associations), and if it can be demonstrated from longitudinal assessments that the exposure preceded the outcome. Since data from observational studies may never completely address residual confounding or precisely identify temporal associations, data often are needed from experimental studies, notably randomized controlled trials, that address whether altering the exposure (i.e., treating OSA) leads to a change in the health outcome. The existing epidemiological literature that has addressed OSA and CVD has been limited by intrinsic challenges in designing the large-scale studies required to quantify associations of modest effects, by the complexities in accurately and reliably measuring OSA in large-scale research studies, and by analytical challenges in
dissecting the role of confounders. This chapter will review the existing database of moderate-to large-scale studies that have addressed the association of OSA to hypertension and CVD.
