Discussion
In this prospective observational cohort study, we have demonstrated that simple spirometric measurements of lung volume and clinic SpO2 predict chronic respiratory failure in obese patients with sleep-disordered breathing. We have also shown, as expected, gender differences for predicting hypercapnic respiratory failure. Surprisingly, previous data have only shown weak correlations between body composition, lung volume and chronic respiratory failure.16 ,17 However, these studies have not considered the clinical physiological predictive power of utilising FVC and daytime clinic oxygen levels (SpO2) in obese patients with confirmed evidence of sleep-disordered breathing to enhance screening for chronic respiratory failure. Indeed, we considered that this is a major caveat of the previous studies, and this is the first study to have shown in a large group of obese patients that a clinic SpO2 level below 95% for a man and 93% for a woman and an FVC less than 3.5 L for a man and 2.3 L for a woman has high sensitivity to detect hypercapnia, albeit with low specificity. Such tests with a high NPV will ensure that few patients with hypercapnia are missed, albeit at the expense of arterial blood gas measurements being performed on a few patients with eucapnia.
Critique of the method
These data were collected from a single centre specialising in the management of patients with sleep-disordered breathing and chronic respiratory failure. Although this was a selected cohort of obese patients with sleep-disordered breathing, unlike previous studies,18 the current study has identified that the simple clinical tests of FVC and SpO2 are not only reduced in patients with chronic respiratory failure, but also FVC and SpO2 can predict hypercapnic respiratory failure. From the current study, all the patients had sleep-disordered breathing, and it is the presence of a restricted lung volume that identifies the patients with chronic respiratory failure. Although the important associations that we have observed do not necessarily imply a causal link between anthropometric parameters, spirometric parameters and hypercapnic respiratory failure in obese patients, the results are physiologically coherent. The simplest method to test this as a causal mechanism between gas exchange, body composition, sleep-disordered breathing and lung volumes would be to undertake detailed measurements of respiratory polysomnography, pulmonary mechanics and lung volumes before and after significant weight loss, such as prebariatric and postbariatric surgery.19 ,20
Although it is well-recognised that vasoconstriction, dark skin and body movements influence transcutaneous oxygen and carbon dioxide measurements,21 ,22 all the physiological measurements were taken according to international and local guidelines and repeated in line with these standards.14 ,23 A simple screening tool based on absolute, rather than % predicted FVC values, would maximise the distribution and application of the test, obviating the need for calculating the % predicted FVC. Therefore, % predicted FVC was not included in the multiple linear regression model and absolute FVC measurements were used instead. As expected, gender differences in height would have affected absolute lung volume measurements, but we developed absolute ‘cut-offs’ for SpO2 and FVC that predict hypercapnia in men and women, respectively. This simple approach to screening would facilitate the incorporation of absolute FVC measurement into clinical practice.
Differences between patients with hypercapnia and eucapnia
To our knowledge, this is the largest UK dataset comparing patients with eucapnia and hypercapnia with sleep-disordered breathing. Interestingly, patients with eucapnia and hypercapnia had similar levels of daytime somnolence demonstrating that the standard clinical assessment, using the ESS,15 does not discriminate between those patients with and without chronic respiratory failure. However, as expected, the clinic SpO2 and PaO2 were lower in the hypercapnic group and SpO2 time below 90% higher, and the relationships between these variables were consistent with previous reports.6 ,24 The patients with chronic respiratory failure had a longer length of stay for establishment of nocturnal respiratory support than those with eucapnia, as a consequence of the extended titration of nocturnal NIV compared with CPAP, which is more straightforward and often can be performed on an outpatient basis.25
Relationship between daytime gas exchange, body composition and lung volume
In patients with sleep-disordered breathing, univariate analysis demonstrated that clinic SpO2 was correlated, albeit weakly, with BMI, FVC and PaCO2. In addition, PaCO2 was weakly correlated with BMI and FVC. These data provide support for the evidence of these weak direct and inverse associations between body composition, lung volume and hypercapnia. More importantly, multivariate linear regression analysis demonstrated that FVC and SpO2 were retained in the model predicting hypercapnia, and the ROC analysis demonstrated that FVC and SpO2 had sufficient sensitivity to be clinically useful predictors of chronic respiratory failure, albeit the specificity was low. These data have shown that an abnormal overnight oximetry study combined with a clinic SpO2 and an FVC measurement could be used as part of a sleep screening programme. The rationale for any future study would be to predict hypercapnic respiratory failure by employing a sleep screening strategy of overnight home oximetry to identify patients with sleep-disordered breathing combined with an FVC and a clinic SpO2 cut-off level. A clinically relevant primary event-related outcome in terms of health service research would be time to diagnose and treat hypercapnic respiratory failure with, secondary end-points including cost utility analysis and patient satisfaction.
Clinical implications
With the increasing incidence of obesity, general practitioners, as well as respiratory sleep clinics, have to be fully cognisant of the respiratory effects of obesity, including patients presenting with chronic respiratory failure. Such patients need to be established on NIV and potential delays and clinical pathway breaches will occur unless a streamlined system is in operation. A number of units already dispatch, via a courier and postal service, an oximeter prior to the clinic appointments.12 In those in whom nocturnal oximetry is available, the addition of simple SpO2 and FVC measurements can be performed by ancillary staff; the clinician can stratify those who would require an arterial blood gas to confirm chronic respiratory failure or can be used as additional verification for borderline oximetry results. Although arterial blood gas measurement and analysis is a relatively straightforward physiological test to perform, it is painful for the patient, requires skill of the clinician to perform and access to costly analysis equipment, which may not be available in all clinic settings. By using simple non-invasive screening tests of FVC and clinic SpO2, hypercapnia can be predicted in obese patients with sleep-disordered breathing prior the clinic appointment and arrangements made for arterial blood gas sampling to be performed either prior to or during a clinic attendance. This would not only limit the number of patients undergoing arterial blood gas measurements, which would clinically benefit the patient and financially benefit the service, but also ensure the most appropriate treatment is delivered to the patient in a timely fashion.