Validation of a Predictive Score for HAST

Over the last two decades, there has been a significant increase in the number of people flying on a regular basis. As a result, more individuals with chronic respiratory disease are being referred for formal assessment to assess if they require inflight oxygen. At cruising altitude, passengers are exposed to reduced atmospheric pressure equivalent to an altitude of 8000 ft, corresponding to a reduction of oxygen from 21% to approximately 15%. National and International recommendations on air travel suggest screening patients with chronic lung diseases for the need of in-flight oxygen supplementation. However, due to the paucity of available evidence, there is no consensus on which investigation should be performed and the precise risk factors associated with in-flight complications. The British Thoracic Society, which published the most recent and comprehensive recommendations on this topic, recommends that patients with respiratory diseases should be assessed clinically, and in specific situations, a hypoxic altitude simulation tests (HAST) should be performed. Unfortunately, the vast majority of these recommendations are level C at best, as current evidence is limited to a handful of prospective or retrospective analyses. A standard hypoxic altitude simulation test requires the patients to breathe a blend of 15% oxygen in nitrogen for 20 minutes. This simulates the oxygen tension at 8000 ft, the equivalent cabin pressure. Blood gas measurements are taken before and after the test. A fall in the pO2 below 6.6 kPa is considered indicative for use of in-flight oxygen. Often the HASTs yield negative results (i.e. no inflight oxygen required). These tests are heavy on resources, time consuming, costly, not widely available and require dedicate hospital appointment. Thus, predictors of outcomes of HASTs are needed to reduce the number of HASTs performed. Several studies have looked at predictors of oxygen desaturation during HAST in varying populations with the aim of better selecting patients who need the test. However, results have been inconclusive. Data on the role of baseline oxygenation, FEV1 and exercise test in predicting oxygen desaturation during the HAST are conflicting and most studies have been too small to provide decisive answers. The Investigators recently developed and validate, on historical data, three scores to predict negative HASTs, according to the BTS 2011 recommendations (i.e. scores predicting which patients do not need to undergo a formal assessment). All of these had good ROC AUC and agreement with the outcome of the HAST in the retrospective dataset. The Investigators hypothesise that the multivariable score (Model 1), based on a combination of CBG variables (bicarbonate and oxygen saturation) and baseline diagnosis (ILD and COPD) would be the best predictor of negative response in a formal HAST. Furthermore, the Investigator hypothesise that the score could be further simplified for the use in clinical practice by measuring the serum bicarbonate and oxygen saturation through a pulse oximeter (Model 1s). This is in light of the good arterio-venous concordance previously demonstrated for bicarbonate and of the good reliability of non-invasive measurement of oxygen saturation. Such a simplified score, if its predictive value is confirmed, would not only reduce the number of unnecessary HAST performed, but also minimise patients' discomfort and hospital appointments, reducing the costs for the NHS. In this prospective single centre study, the Investigators will assess the validity of the best performing score (Model 1) derived in our retrospective study for negative HAST. In addition, the Investigators want to verify whether the use of SpO2 in place of SO2 and of serum bicarbonate in place of capHCO3 would allow for similar results. The investigators also aim to evaluate the performance of the other predictive scores previously derived in our retrospective analysis, and, if none of these models are a good and effective predictor, we aim to develop a new model based on newly measured data. The Investigators aim to enroll up to 280 subjects, but interim analyses will be performed with the scope to early terminate the study if primary outcome is met. Demographic and baseline characteristics of patients will be analysed on the whole population and by the cohorts depending on the underlying diagnosis and the outcome of the HAST. Receiver operating characteristics (ROC) curves will be plotted to test the hypothesis that the scores are good predictors of negative HASTs. Diagnostic values of the scores will be expressed by sensitivity, specificity, positive and negative predictive values.