In clinical practice discrepancies between overnight SpO2 recordings performed by 2 devices used simultaneously are regularly observed. However, this has not been systematically studied or quantified. It is therefore important to determine if these discrepancies are anecdotic, or frequent, and to what extent this may affect decisions in clinical practice such as implementing (or withdrawing) oxygen in subjects under noninvasive ventilation (NIV) or continuous positive airway pressure (CPAP), or adjusting NIV settings.

Pulse oximetry estimates oxygen saturation in the arterial blood by trans-illuminating a translucent tissue (usually a fingertip, or an ear lobe) with light-emitting diodes at 2 specific wavelengths. Absorption of light at these different wavelengths (660 nm, red, and 940 nm, infrared respectively) differs between oxygenated and deoxygenated hemoglobin. The amount of light which is transmitted at both wave lengths is then quantified and processed by an algorithm that displays a saturation value. The signals are corrected for the pulsatile nature of arterial oxygen flow. Devices also provide pulse rate measured by plethysmography (pulse-related changes in volume of fingertip or earlobe). Since its introduction in the early 1980s, pulse oximetry has proven to be an essential tool for the non-invasive assessment of blood oxygen saturation (SpO2). The use of pulse oximeter is now widespread and interests acute care settings as well as primary care settings. Although there have been recent improvements in signal analysis such as increasing sampling frequency and improving reflectance technology [4], the accuracy of commercially available oximeters differs. First of all, there is variability in the way accuracy of pulse oximetry devices is reported within 2% (± 1 SD) or within 5% (± 2 SD) of reference measurements obtained by blood gases analysis [5]. Secondly, there is a variability between commercially available devices, especially below a SpO2 of 90%. This is partly explained by the fact that calibration of the different algorithms employed in signal processing is limited by the range of saturations that can be safely obtained in healthy volunteers. In respiratory medicine, it has been shown that such variability of accuracy could affect the diagnosis of Obstructive Sleep Apnea Hypopnea Syndrome (OSAHS) and impact on clinical decisions as the recorded number of apneas/hypopneas varied between devices during nocturnal sleep studies [8] Nocturnal hypoxemia (NH) is considered as one of the major determinants of adverse cardiovascular complications and neurocognitive impairment in patients suffering of chronic respiratory failure (CRF). Because of its simplicity, short set-up time and short time response, pulse oximetry is a valuable screening tool for nocturnal hypoxemia despite its disadvantages such as motion artefacts or sensitivity to perfusion. Therefore, definitions of NH rely solely on nocturnal oximetry recording: for instance, in a consensus statement on noninvasive ventilation, spending > 5% of sleep time under 88% of SpO2 was considered as a relevant threshold. Definitions of NH remain arbitrary and different expert-based thresholds have been suggested in the medical literature [10]. Patients suffering from nocturnal hypoventilation, especially those with an average daytime SpO2 close to the steep portion of the hemoglobin dissociation curve (SpO2 between 90-94%), are at higher risk for NH. Therefore, in these patients, device imprecision could have a significant impact on medical decisions, such as deciding to adjust ventilator settings in patients under noninvasive ventilation (NIV) and/or implementing nocturnal oxygen supplementation.

In patients with CRF and NIV, after optimal adjustment of ventilator settings, prescription of nocturnal oxygen supplementation is common practice although impact of nocturnal oxygen supplementation on survival, patient comfort, or prevention of cor pulmonale has yet to be demonstrated. Practically speaking, it increases considerably the burden of the treatment for the patient and care givers (additional connections and tubings, noise of the oxygen concentrator etc.). To our knowledge, no study has evaluated how the use of different pulse oximeters could impact on this decision. Three types of devices are used in clinical practice: pulse oximetry using a probe connected to the home ventilator device, pulse oximetry combined with transcutaneous capnography, or using a "wrist watch" type of device.

• Inclusion criteria:

  • > 18 years old
  • Awake room air SpO2 between 90 and 94%
  • NIV or CPAP therapy
  • Ambulatory or hospitalized patient in a clinically stable respiratory condition without any vasopressor treatment.

• Exclusion criteria:

  • Hospitalization in an acute care setting (e.g. emergency room, Intensive Care Unit, Intermediate Care Unit)
  • Any vasopressor treatment
  • Peripheral vascular pathologies that can affect digital perfusion (e.g. history of ischemia, Raynaud's phenomenon, any type of vasculitis).
  • Mechanical obstacles that may limit quality of signal (e.g. nail polish, bandage, splint, plaster).
  • Patient already treated by long term nocturnal oxygen therapy (LTOT)

Participants will be recruited among 1) hospitalized patients with NIV treatment; 2) ambulatory patients followed for NIV therapy at the outpatient unit of the Division of Pulmonology; 3) patients undergoing an elective polysomnography (sleep study) for adjustment or titration of a positive pressure device (NIV or CPAP) at the Sleep Laboratory of the Division of Pulmonology.