213 Summary and general discussion 9 were performed in the RECOVERY trial, it was estimated that 494 participants would be needed in each study arm. In conclusion, S/F94 is superior to S/F as a measure of pulmonary oxygenation function and is an effective intermediate outcome measure in COVID-19. Comparing multiple widely used outcome measures, fewer patients would be needed to detect the same treatment effect when using S/F94. S/F94 is a simple and non-invasive measurement, representative of disease severity and provides greater statistical power to detect treatment differences than other intermediate endpoints. In Chapter 8, different sub-phenotypes of patients with Staphylococcus aureus bacteraemia (SAB) were identified. SAB is a clinically heterogeneous disease, and many previous clinical trials failed to show a treatment effect52. It is possible that because the heterogeneity, we are unable to identify subgroups of patients who benefit (or suffer harm) from a specific treatment. Over 1400 adult patients from three different cohorts were included in the analysis: a U.K. retrospective observational study (Edinburgh cohort, n=458), the U.K. multi-centre ARREST randomised trial50 (n=758), and the Spanish multi-centre SAFO randomised trial52 (n=214). Participants of the ARREST trial were randomised to receive either adjunctive rifampicin or placebo in addition to standard antibiotic treatment. Participants of the SAFO trials were randomised to receive cloxacillin plus Fosfomycin or cloxacillin alone. Latent class analysis (LCA) was used to identify sub-phenotypes. A sub-phenotype is considered a subgroup of patients who have similar characteristics. Using clinical indicator variables, homogenous sub-phenotypes were identified within the larger, heterogenous cohort of SAB patients. The classes were formed without consideration of any clinical outcomes. Model selection was done based on a combination of clinical interpretation, the Bayesian Information Criteria (BIC), the number of classes and the size of the smallest class. After identification of the optimal number of classes, the posterior probability of class membership for each of the identified classes was estimated for each individual. Each individual was then assigned to the class with the highest probability, using the bias-adjusted three-step LCA for misclassification errors53. Z-scores were calculated to compare class-defining variables between sub-phenotypes. Eighteen class-defining variables were included for the LCA. For both the Edinburgh and ARREST cohorts, a 5-class model was considered best. Sub-phenotype A was associated with older age, co-morbidity and SAB from unknown or skin of soft tissue infection source. Metastatic foci were relatively uncommon in this subphenotype. Sub-phenotype B was associated with nosocomial SAB, younger age, fewer co-morbidities, bacteraemia originating from an intravenous catheter and lack of any metastatic foci. Sub-phenotype C was associated with community-acquired SAB from unknown source, a higher C- reactive protein (CRP) and metastatic foci of infection. Sub-phenotype D was associated with chronic kidney disease (CKD), intravenous catheter source and nosocomial or healthcare associated acquisition.
RkJQdWJsaXNoZXIy MTk4NDMw