We are grateful to Simpson and colleagues for their comments on our article. They suggest that we should have only examined the effect of habitual snoring, implying that all snorers are a large heterogeneous group unlikely to be associated with adverse clinical outcomes. In fact, our data allowed us to test this hypothesis and we did show some evidence of ordinal associations with high density lipoprotein (HDL) cholesterol and with pulse wave velocity (in children below the median body mass index; see table 2), although there was no evidence of other associations. We believe that our ordinal scale analysis is more powerful than the dichotomous classification proposed by Simpson et al. as the definition of habitual snoring (three nights or more a week) is arbitrary and dichotomisation at the wrong point in this scale introduces the possibility of missing something important.
Drawing inferences about the causality based on observational studies in clinic-based samples is hazardous.1 We agree that in children, severe obstructive sleep apnea (OSA) confirmed by polysomnography (PSG) probably does affect systemic blood pressure. This conclusion is based on evidence from community-based cross-sectional studies from Hong Kong, n= 190,2 Pennsylvania, n= 7003 and Tuscon, n= 2394 combined with evidence from studies conducted in adults and in animals. However, we do not find convincing evidence that snoring alone causes hypertension, and there do not seem to be community-based data concerning other cardiometabolic factors. The Hong Kong group found that only one of 22 blood pressure outcomes, nocturnal diastolic blood pressure, was elevated in primary snorers. The additional trend analyses in that paper, which also included two additional categories of definite sleep apnoea, do not allow one to conclude that snoring without sleep apnoea causes hypertension. The study from Pennsylvania found no association between snoring and blood pressure in those without sleep apnoea. In the mild sleep apnoea group, they did observe an effect on blood pressure (the moderate OSA group contained on seven children from 700 making effect modification impossible to gauge). We look forward to seeing these results replicated elsewhere along with our observation here and in adults of a potential causal relationship between sleep-disordered breathing and HDL-C levels.5
Our study did not employ PSG and therefore did not seek to assess the impact of sleep apnoea, as we acknowledged in the manuscript. On the other hand, a strength of this study was the wide range of objective cardiometabolic risk factor measurements we made in a relatively large number of children. Our findings support the view of Simpson et al. that treatment decisions should be made only after a sleep study has confirmed the presence of sleep apnoea.