The Burden of Proof Lies with the Prosecution: Is Snoring Guilty?

Abstract

615 Letter to the Editor—Marshall et al We are in broad agreement with Drs. Kezirian and Chang’s letter1 regarding our publication describing no statistically discernible increase in the risk for death, incident cardiovascular disease, or stroke that can be linked to snoring over 17 years observation in the Busselton Sleep Cohort.2 They have quite rightly highlighted the fact that as a small cohort study, we were unable to rule out the possibility that snoring has subtle effects on cardiovascular health because our confidence intervals were quite wide. But as our editorialist noted,3 there do not appear to be any other community-based cohort studies with sufficient follow-up that might provide additional high-quality data. So even with all of the problems highlighted by us2 and by the letter,1 our study seems to be the only one of its type. The study by Rich and colleagues did report a subtle association between a measured snoring index and all-cause mortality, but only in a subset of patients (n = 5,955) without sleep apnea and with a BMI < 30 (OR = 1.16; 95% CI, 1.01-1.32) when the background mortality rate in the whole cohort was 2.1% in n = 77,260 patients.4 But that study was not a community-based cohort, and the authors regretted being unable to control for basic risk factors such as cholesterol and smoking. And so we must be left wondering whether those explain the risk in that subgroup analysis. Because of the relatively small size of our study we chose to provide the power we had to detect quite large increases in the relative risk (i.e., 3 times the risk). Whilst some cardiovascular risk factors have a relatively small relative increase in risk, they tend to have important increases in absolute risk for bad outcomes if the risk factor and the outcome are common—but the stroke outcome in our study was quite rare. We still agree with Kezirian and Chang that the Busselton cohort shows that there probably is not a strong association, but there may be a weak association (i.e., an odds ratio of 1.2-1.5 between the lowest and highest quartile of snoring). As we stated in the paper,2 we were severely underpowered to detect subtle associations, particularly with a rare occurrence such as the biologically plausible stroke association. To borrow their legal metaphor: in both law and medicine the burden of proof is on the accuser. So far there is still very little reliable evidence from studies of the right scientific design that snoring is a killer. The fatal snoring hypothesis seems unlikely to us given the difficulty in showing that mild sleep apnea deleterious consequences.5-8 In addition, the influential Spanish LETTER TO THE EDITOR