<p>Oxygen-induced elevations in arterial carbon dioxide tension have been demonstrated in patients with chronic obstructive pulmonary disease (COPD), asthma, pneumonia, obesity hypoventilation syndrome (OHS) and acute lung injury. A randomised controlled trial (RCT) in acute exacerbations of COPD (AECOPD) found an over two-fold increase in mortality in patients randomised to high concentration oxygen, compared to titrated oxygen. These findings support guideline recommendations for titration of oxygen therapy to a target oxygen saturation range, reducing the risks of hypoxaemia and hyperoxaemia. This thesis focuses on the potential implications of oxygen-induced elevations in carbon dioxide in the acute clinical setting. The reviews and studies in the following chapters are all aimed at addressing gaps in knowledge which may have practical implications for oxygen therapy and/or the identification of patients at risk of oxygen-induced hypercapnia in clinical practice. Numerous studies have demonstrated that high concentration oxygen continues to be administered to acutely unwell patients, despite guideline recommendations for titrated therapy. The first study in this thesis is a clinical audit evaluating the effects of a staff education program, which included face-to-face and written training for ambulance staff. The education program was associated with reduced the rates of high concentration oxygen administration to patients with AECOPD. This suggests active education may increase adherence to oxygen guidelines among clinical staff. The ability to avoid hypoxaemia and hyperoxaemia during titrated oxygen therapy relies on appropriate lower and upper target oxygen saturation limits, which may be impacted on by pulse oximeter accuracy. The second study in this thesis is a multicentre observational study in which 400 paired pulse oximeter (SpO₂) and arterial blood gas saturation (SaO₂) values were collected in the hospital setting. A SpO₂ <92% had 100% sensitivity for detecting SaO₂<90%. This indicates guideline recommended target oxygen saturations of 92-96% adequately avoid hypoxaemia. Two studies in OHS patients have investigated the effects of oxygen administration on carbon dioxide, however their designs, including recruitment of stable participants, have limited their generalisability to clinical practice. Therefore, a cross over RCT was conducted in 24 morbidly obese hospital inpatients, randomised to the order they received high concentration and titrated oxygen, each for 60 minutes. The mean change in the transcutaneous partial pressure of carbon dioxide (PtCO₂) from baseline was 3.2 mmHg higher during high concentration oxygen, compared with titrated oxygen (P=0.002). This supports guideline recommendations to titrate oxygen in patients with obesity, regardless of whether they have a diagnosis of OHS or not. The effects of oxygen in patients with bronchiectasis, neuromuscular disease or kyphoscoliosis are uncertain. Stable patients with these conditions were recruited to double-blind randomised cross over trials administering air and 50% oxygen, each for 30 minutes. A trial was also performed in stable COPD patients for comparison. There was no significant change in PtCO₂ with oxygen therapy in the neuromuscular disease/kyphoscoliosis patients. In the bronchiectasis and COPD patients, oxygen was associated with increased PtCO₂ from baseline compared to air, but the differences were not clinically significant (0.4 mmHg, P=0.012 and 1.3 mmHg, P<0.001, respectively). The lack of a clinically significant PtCO₂ increase in the COPD patients indicated the study findings were unlikely to be generalisable to the clinical setting, and highlights the potential limitations in applying data from stable participants to patients who require acute oxygen therapy. These studies support current guideline recommendations for titrated oxygen therapy, provide insight into the limits of studying the effects of oxygen in stable participants, and demonstrate the utility of an educational program to aid the translation of research findings into relevant changes in clinical practice.</p>
Model of SpO2 signal of the neonate