ABSTRACT
Optimising statistical power in early-stage trials and observational studies accelerates discovery and improves the reliability of results. Ideally, intermediate outcomes should be continuously distributed and lie on the causal pathway between an intervention and a definitive outcome such as mortality. In order to optimise power for an intermediate outcome in the RECOVERY trial, we devised and evaluated a modification to a simple, pragmatic measure of oxygenation function - the SaO2/FIO2 (S/F) ratio. We demonstrate that, because of the ceiling effect in oxyhaemoglobin saturation, S/F ceases to reflect pulmonary oxygenation function at high values of SaO2. Using synthetic and real data, we found that the correlation of S/F with a gold standard (PaO2/FIO2, P/F ratio) improved substantially when measurements with SaO2 [≥] 0.94 are excluded (Spearman r, synthetic data: S/F : 0.31; S/F94: 0.85). We refer to this measure as S/F94. In order to test the underlying assumptions and validity of S/F94 as a predictor of a definitive outcome (mortality), we collected an observational dataset including over 39,000 hospitalised patients with COVID-19 in the ISARIC4C study. We first demonstrated that S/F94 is predictive of mortality in COVID-19. We then compared the sample sizes required for trials using different outcome measures (S/F94, the WHO ordinal scale, sustained improvement at day 28 and mortality at day 28) ensuring comparable effect sizes. The smallest sample size was needed when S/F94 on day 5 was used as an outcome measure. To facilitate future study design, we provide an online user interface to quantify real-world power for a range of outcomes and inclusion criteria, using a synthetic dataset retaining the population-level clinical associations in real data accrued in ISARIC4C https://isaric4c.net/endpoints. We demonstrated that S/F94 is superior to S/F as a measure of pulmonary oxygenation function and is an effective intermediate outcome measure in COVID-19. It is a simple and non-invasive measurement, representative of disease severity and provides greater statistical power to detect treatment differences than other intermediate endpoints.
ABSTRACT
BackgroundRapid and accurate detection of SARS-CoV-2 infected individuals is crucial for taking timely measures and minimizing the risk of further SARS-CoV-2 spread. We aimed to assess the accuracy of exhaled breath analysis by electronic nose (eNose) for the discrimination between individuals with and without a SARS-CoV-2 infection. MethodsThis was a prospective real-world study of individuals presenting to public test facility for SARS-CoV-2 detection by molecular amplification tests (TMA or RT-PCR). After sampling of a combined throat/nasopharyngeal swab, breath profiles were obtained using a cloud-connected eNose. Data-analysis involved advanced signal processing and statistics based on independent t-tests followed by linear discriminant and ROC analysis. Data from the training set were tested in a validation, a replication and an asymptomatic set. FindingsFor the analysis 4510 individuals were available. In the training set (35 individuals with; 869 without SARS-CoV-2), the eNose sensors were combined into a composite biomarker with a ROC-AUC of 0.947 (CI:0.928-0.967). These results were confirmed in the validation set (0.957; CI:0.942-0.971, n=904) and externally validated in the replication set (0.937; CI:0.926-0.947, n=1948) and the asymptomatic set (0.909; CI:0.879-0.938, n=754). Selecting a cut-off value of 0.30 in the training set resulted in a sensitivity/specificity of 100/78, >99/84, 98/82% in the validation, replication and asymptomatic set, respectively. InterpretationeNose represents a quick and non-invasive method to reliably rule out SARS-CoV-2 infection in public health test facilities and can be used as a screening test to define who needs an additional confirmation test. FundingMinistry of Health, Welfare and Sport Research in contextO_ST_ABSEvidence before this studyC_ST_ABSElectronic nose technology is an emerging diagnostic tool for diagnosis and phenotyping of a wide variety of diseases, including inflammatory respiratory diseases, lung cancer, and infections. As of Feb 13, 2021, our search of PubMed using keywords "COVID-19" OR "SARS-CoV-2" AND "eNose" OR "electronic nose" OR "exhaled breath analysis" yielded 4 articles (1-4) that have assessed test characteristics of electronic nose to diagnose COVID-19. In these small studies the obtained signals using sensor-based technologies, two-dimensional gas chromatography and time-of-flight mass spectrometry, or proton transfer reaction time-of-flight mass spectrometry, provided adequate discrimination between patients with and without COVID-19. Added value of this studyWe prospectively studied the accuracy of exhaled breath analysis by electronic nose (eNose) to diagnose or rule out a SARS-CoV-2 infection in individuals with and without symptoms presenting to a public test facility. In the training set with 904 individuals, the eNose sensors were combined into a composite biomarker with a ROC-AUC of 0.948. In three independent validation cohorts of 3606 individuals in total, eNose was able to reliably rule out SARS-CoV-2 infection in 70-75% of individuals, with a sensitivity ranging between 98-100%, and a specificity between 78-84%. No association was found between cycle thresholds values, as semi-quantitative measure of viral load, and eNose variables. Implications of all the available evidenceThe available findings, including those from our study, support the use of eNose technology to distinguish between individuals with and without a SARS-CoV-2 infection with high accuracy. Exhaled breath analysis by eNose represents a quick and non-invasive method to reliably rule out a SARS-CoV-2 infection in public health test facilities. The results can be made available within seconds and can therefore be used as screening instrument. The eNose can reliably rule out a SARS-CoV-2 infection, eliminating the need for additional time-consuming, stressful, and expensive diagnostic tests in the majority of individuals.
