182 Chapter 8 The Infectious Diseases Society of America (IDSA)26 and Fowler et al27 definitions of complicated SAB were applied to the Edinburgh cohort, identifying a lower proportion of patients meeting both definitions of complicated SAB in sub-phenotype B (nosocomial intravenous catheter SAB) and a higher proportion in C (communityacquired metastatic SAB; Supplementary Figure 3A). Two current definitions of ‘low risk’ SAB were also applied to the Edinburgh cohort (the SABATO trial eligibility criteria28,29 and the definition used by Hendriks et al30), with patients meeting these definitions predominantly predicted to belong to sub-phenotype B (Supplementary Figures 3B and C). The distribution of spa type inferred clonal complexes (Supplementary Figure 3D) did not differ substantially between sub-phenotypes in the Edinburgh cohort. Predicted members of sub-phenotypes A (older co-morbid SAB) and D (CKD SAB) in the Edinburgh cohort had the highest Charlson Comorbidity Index, with the predicted members of sub-phenotypes B and E having the lowest (Supplementary Figure 3E). The SAFO trial included a substantially smaller number of participants than the Edinburgh and ARREST cohorts, applied more stringent inclusion/exclusion criteria (excluding MRSA infection, moderate-severe heart failure, and IDU), and did not record baseline CRP. Combining model fit parameters and interpretability, a fourclass model was favoured. As expected, sub-phenotype E (IDU SAB) was not identified but the other four classes identified were similar to sub-phenotypes A-D identified in the Edinburgh and ARREST cohorts (Figure 1). SAFO sub-phenotype A was associated with older age, vascular disease, and SAB originating from SSTI. SAFO sub-phenotype B was associated with younger patients with nosocomial SAB from intravenous catheter source. SAFO sub-phenotype C was associated with community-acquired metastatic SAB from an unknown source. SAFO sub-phenotype D was associated with healthcare-associated SAB, chronic kidney disease, and intravenous catheter source. Predicted members of SAFO sub-phenotypes A and D had the highest Charlson Comorbidity Index, and B had the lowest, consistent with the associations seen in the Edinburgh cohort (Supplementary Figure 4). Clinical outcomes of SAB sub-phenotypes Differences in 84-day mortality and microbiologic outcomes were observed between the sub-phenotypes (Figure 2A; Supplementary Table 1). In both the Edinburgh cohort and ARREST placebo arm (n=388), 84-day mortality was highest in sub-phenotype A and lowest in sub-phenotypes E and B. In the Edinburgh cohort, people assigned to sub-phenotypes A and D had the lowest one-year survival, whereas those assigned to E had the highest (Supplementary Figure 5). In the Edinburgh cohort, sub-phenotype C was associated with increased rates of persistent or recurrent bacteraemia (Figure 2B). In the ARREST placebo arm cohort, sub-phenotype C was also associated with a higher rate of composite microbiologic failure (Figure 2C). In both cohorts, sub-phenotype B was associated with lower rates of microbiologic failure. The smaller number of patients in the SAFO control arm (n=110) limited our ability to compare outcomes between the sub-phenotypes, but similar patterns were observed (Supplementary Figure 6). People assigned to sub-phenotypes A and D
RkJQdWJsaXNoZXIy MTk4NDMw