Sleep disordered breathing is such a common clinical problem. We do many diagnostic polysomnograms (PSGS) each year around the world. Should we be collecting more information as part of the testing process to guide decision making? In this issue, Brillante et al. report on an intriguing retrospective study which suggests that an overnight increase in carbon dioxide levels (>7mm Hg) and evening hypoxemia (PO2 < 65mm Hg) preceding a diagnostic polysomnogram (PSG) predicted long-term mortality in sleep disordered breathing.1
It is likely to be a unique study because it is backed by an unusual clinical practice, the collection of evening and early morning arterial blood gases (ABGS) in nearly all patients presenting to the Sleep Laboratory for a diagnostic PSG.
While the study has some limitations, there was nevertheless some very important data that give us much to think about.
Why would what seems a modest change in CO2 levels be associated with an increase in mortality?
At first pass, it does seem likely that obesity hypoventilation syndrome (OHS) patients account for some of the increased mortality. While no long-term randomized controlled trials exist, it seems patients with OHS do seem to have a higher morality risk. Nowbar et al. showed 18 months after hospital discharge, mortality was 23% in a group of patients with OHS compared with 9% in those with uncomplicated obesity.2
However in the Brillante et al. study, only 3 of 21 patients in this series who had a rise in CO2 > 7 mm Hg overnight met strict criteria for OHS that is,
- • Obesity (BMI >30 kg/m2)
- • Awake alveolar hypoventilation (PaCO2 >45 mm Hg)
- • An alternative cause of the hypoventilation cannot be identified.3
In Table 6, the authors identify 31 patients met the criteria for OHS. Intriguingly, only three of these had a CO2 >7 mm Hg overnight. This raises the possibility that it may be that fluctuations in CO2 rather than chronic hypercapnia that puts patients at greater risk. It could however be that OHS was a contributor to the evening hypoxemia results, and the number of patients meeting OHS criteria in this group is not quantified.
Other aetiologies for sleep hypoventilation, for example, COPD may explain some of the mortality as 71/322 (22%) of this cohort had COPD. This might explain some of the late evening hypoxemia that was associated with an increase in mortality. Only a smaller number reported having available spirometry (17/322) so it was hard to quantify the severity of COPD in this patient group.
The authors did not have data to describe treatment decisions after diagnosis. We are told it was policy of the Unit to prescribe continuous positive airway pressure (CPAP) therapy in most of these clinical circumstances however, so it seems reasonable to assume a proportion of these patients went onto CPAP therapy, but there is no way of confirming this. Furthermore, given that nocturnal hypoxemia predicted mortality, did these patients' oxygen saturations improve and normalize on therapy? Should more have been started on oxygen therapy either with positive pressure ventilation support or as a stand-alone therapy? It is not clear how many may have started bi-level ventilation, if any. All of this raises three intriguing possibilities: (i) mortality would have been higher if CPAP was not used; or (ii) CPAP therapy had no impact on mortality in this patient group; or (iii) more of these patients should have been on bi-level ventilation rather than CPAP therapy. Given the recent RCT reported by Piper et al. in which it was reported that for patients with OHS randomized to either CPAP or bi-level therapy for a three-month period, no between-group differences in awake CO2, compliance with therapy or daytime sleepiness were found; the third possibility seems less likely to be the key issue.4
It is not clear if the clinical management was changed by the results of the evening ABGS. If they were, this may contribute to some of the signal here. For example, if patients had a low PO2 reading in the evening is it possible, probably even likely, that some would be placed on nocturnal oxygen during their PSG? It is well described that sleep hypoventilation may occur due to increased nocturnal oxygen flow in hypercapnic COPD patients.5 Could it be therefore that some of the rise in PCO2 followed acute oxygen administration?
There are limitations to the study of course. Data analysis was retrospective. It seems surprising that while this data series is called consecutive, so few patients had data available for review as 322 patients represent only 1.3 patients/week. The authors describe the reasons for this in Figure 1.1
The rise in PCO2 of >7 mm Hg overnight is not a measurement used clinically, and even if it was, I think it is unlikely that it would be hard to justify routine ABGS on all patients to look for this rise. Should all patients with low resting saturations have a PM and an AM arterial blood gas as the authors suggest? A single arterial blood gas will not detect change in PCO2 after all. Unfortunately, non-invasive methods of measuring PCO2 remain notoriously unreliable and unstable and thus would be unlikely to pick up such subtle changes in PCO2 across the night.6
The authors are to be commended on a thoughtful and clever review of the data they had available to them. While questions remain, it is worthwhile that we consider impact of hypoxemia and the change in carbon dioxide levels more carefully as we review the many patients who are assessed for sleep disordered breathing in 2012.Those with these abnormalities may be the ones at greatest risk, and it seems highly unlikely that Australia's proposed carbon tax will have any impact on these carbon dioxide levels.