BAC DOOR: A Clinician Ranking Exercise for Better S. aureus Bacteremia Trial Design

Background: New therapies are needed for S. aureus bacteremia (SAB). However, when comparing new antibiotic regimens, the oft-used noninferiority design fails to address the fundamental question of which treatment is better for patients. Desirability of Outcome Ranking (DOOR) is a novel method for analysis of clinical trials. Patients are ranked according to overall outcomes, taking into account both benefits and harms. We conducted a study to develop a novel DOOR strategy to utilize in future SAB treatment trials.

Methods: Twenty SAB patient profiles were constructed by a team of ID physicians to represent the range of experiences and outcomes observed in prior trials. The profiles described the efficacy, adverse events (AEs), and treatment adjustments of a patient during a theoretical trial comparing two treatments. A computerized survey with profiles presented in random order was sent to 43 ID clinicians working in the USA (28% pediatric). The respondents were asked to rank the 20 profiles from best to worst on the basis of global outcome. We measured the agreement between respondent ranks and several pre-specified DOOR algorithms developed by the research team. Additionally, a DOOR strategy based on the respondent consensus was developed using classification and regression tree (CART) and other analyses.

Results: Forty-two (97%) respondents completed the survey. Respondent rankings demonstrated good agreement (median Spearman correlation r=0.69, IQR 0.60-0.77). Patients with extreme positive or negative outcomes proved easy to discriminate. However, groups of patients in the middle had similar rankings (Figure 1). CART analyses suggested that features best discriminating rank are survival, severe AE, cure, infectious complications, and ongoing symptoms. Based on these analyses, we developed a DOOR strategy (Figure 2), which correlates well with the respondent consensus ranking (r=0.89).

Conclusion: When comparing SAB patient profiles, respondents place value not just on cure, but also on AEs and symptom resolution. A DOOR strategy incorporating these outcomes can be used for future trials comparing treatment strategies for SAB and other infections with the goal of improved differentiation between management strategies.

Figure  SEQ Figure \* ARABIC 1: Clinician Rankings by Patient