We identified 362 hospital encounters with MRSA-B during the study period, of which 149 encounters were excluded (Fig. 1). The most common reasons for exclusion were polymicrobial bacteremia, culture and/or treatment for MRSA-B at an outside facility prior to transfer, and prior or repeat episodes of MRSA-B within the study period. Therefore, 213 patients met inclusion criteria, of whom 118 patients received monotherapy and 95 patients received combination therapy (Fig. 1). ID consult is mandatory for MRSA-B at our institution, which in practice translated to the involvement of an ID physician in the care of 99.1% (211/213) of patients included in this study.

In the unweighted analysis, patients were older in the monotherapy group (mean 62.96 ± 1.28 years) when compared to the combination therapy group (mean 62.96 ± 1.28 vs. 58.66 ± 1.49 years, P = 0.03) (Table 1). More patients in the combination therapy group had a qPitt score ≥ 2 (49.5% vs. 35.6%, P = 0.04) as well as SBP < 90 mmHg/vasopressor use (44.2% vs. 22.9%, P < 0.01) and RR > 25 per min/requiring mechanical ventilation (66.3% vs. 50.8%, P = 0.02) within 24 h of index blood culture and were more likely to have a respiratory focus of infection. The most common foci of infection were skin and soft tissue infection/surgical and endovascular for both groups, although there significantly more patients in the combination therapy group with a respiratory focus of infection (32.6% vs. 12.7%, P < 0.01). Source control (including partial source control) was achieved in 89% of patients overall. Initial anti-MRSA antibiotics were started significantly faster in the combination therapy group (mean 4.75 ± 2.27 vs. 10.44 ± 1.42 h, P = 0.04). Following incorporation of propensity score weights, there were no significant differences between the two treatment arms.

Twelve patients in the study were found to be COVID positive during their hospitalization for MRSA-B. The overall incidence of COVID in the study was 5.6% (12/213), but was 9.9% (12/121) for patients admitted after 2019. By treatment group, 1 patient (0.85%) was COVID positive in the monotherapy group and 11 patients (11.6%) were positive in the combination therapy group.

Of the 118 patients in the monotherapy group, 115 patients initially received vancomycin and 3 patients received daptomycin. Of these patients, 21.2% of them received monotherapy after the March 2020 consensus statement. The mean time from index blood culture collection to initial anti-MRSA antibiotic was 10.4 h, with 10 (8.5%) patients receiving an anti-MRSA antibiotic before index culture collection (Table 1). Fifteen patients (13%) were escalated to combination therapy with ceftaroline at a mean of 5.4 days. Six of these 15 patients were subsequently de-escalated to monotherapy.

Of the 95 patients who received ceftaroline-based combination therapy, 76 patients initially received combination with vancomycin and 19 patients received combination with daptomycin. Of these patients, 97.9% of them received combination therapy after the March 2020 consensus statement. The mean time to the first anti-MRSA antibiotic was 4.8 h with 13 (13.7%) patients receiving an anti-MRSA antibiotic before index blood culture collection (Table 1). The mean time from MRSA result identified by rapid diagnostic assay to the start of combination therapy was 12.1 h. The mean time from index blood culture collection to the start of combination therapy was 33.2 h. Nineteen patients (26%) who initially received vancomycin plus ceftaroline were switched to daptomycin plus ceftaroline after a mean of 5.8 days. The mean duration of combination therapy was 15.2 days, while the median duration was 8 days.

There were 110 patients who received combination therapy at any point during their treatment course and 70 (64%) were de-escalated to monotherapy (6/15 in the monotherapy group and 64/95 in the combination therapy group). The mean time to de-escalation was 10.4 days. The most common reasons for de-escalation to monotherapy included blood culture clearance with source control and clinical improvement (n = 41), consolidation of regimen at discharge (n = 15), and cytopenia (n = 2). Other reasons included cost, deep-seated infection ruled out, and switch to alternative antibiotics.

In the unweighted analysis, 29 (24.5%) patients in the monotherapy group and 41 (43%) in the combination group met the primary composite outcome of persistent bacteremia, 30-day all-cause mortality, and 30-day bacteremia recurrence (OR 2.23 for combination therapy, 95% CI 1.24, 4.01) (Table 2). When propensity weights were assigned, there was no difference between groups for the primary composite outcome (OR 1.58 for combination therapy, 95% CI 0.60, 4.18). Among the component parts of the propensity-weighted primary outcome, there were greater odds of persistent bacteremia for combination therapy patients (OR 4.28, 95% CI 1.64, 11.19) and no difference between groups for 30-day mortality (OR 1.00, 95% CI 0.33, 3.08) or 30-day recurrence (too few patients to calculate an OR) (Table 2).

At 3 days after the start of therapy, 49 (52%) patients and 91 (77%) patients in the unweighted combination therapy and monotherapy groups, respectively, had negative blood cultures. In the propensity-weighted analysis, time to microbiological cure was longer in the combination therapy group (mean difference 1.50 days, 95% CI 0.60, 2.41) as was hospital length of stay (mean difference 13.38 days, 95% CI 2.43, 24.34) (Table 3). A post-hoc sensitivity analysis excluding 12 outlier admissions with LOS 1.5 times the inter quartile range above the 3rd quartile continued to show a statistically significant longer LOS for combination therapy (mean difference 3.49, 95% CI 0.05, 6.93). Thirty-day readmission was not different between groups in the propensity score-weighted logistic regression (OR 0.74, 95% CI 0.35, 1.54).

We attempted to identify subgroups of patients for which combination therapy may improve outcomes in comparison to monotherapy, hypothesizing that those with greater risk of adverse outcome may show benefit from combination therapy. A high-risk subgroup analysis was performed for patients with a qPitt score > 2 within 24-hours of index culture draw. qPitt score was excluded from the propensity weighting model for this analysis. For the primary composite outcome, this demonstrated an unweighted OR 2.15 (95% CI 0.92, 5.04), and continued to show no difference between groups when propensity weights were assigned (OR 1.09, 95% CI 0.25, 4.74). Among the component parts of the primary outcome, there were greater odds of persistent bacteremia (OR 7.02, 95% CI 1.61, 30.61) and no difference between groups for 30-day mortality (OR 0.58, 95% CI 0.13, 2.66) or 30-day recurrence (too few patients to calculate an OR). There was a significantly longer hospital length of stay (mean difference 23.63 days, 95% CI 1.24, 46.02) for combination therapy patients in this group and no difference between time to microbiologic cure (mean difference 1.08 days, 95% CI -0.50, 2.65) or 30-day readmission (OR 0.95, 95% CI 0.27, 3.35). High-risk subgroup analyses were also performed for patients aged > 65 and patients with CCI > 4 which did not demonstrate any difference in the primary composite outcomes comparing combination therapy and monotherapy.

Safety outcomes were assessed for patients who had available data. Overall, there were no significant differences between the groups (Table 4). The most common ADE was AKI in both groups with 28/71 patients (39%) in the combination therapy group and 38/104 patients (37%) in the monotherapy group experiencing AKI (OR 1.49, 95% CI 0.60, 3.34). More patients demonstrated KDIGO stage 3 AKI in the combination therapy group (OR 2.69, 95% CI 0.60, 12.13). Eosinophilia also occurred slightly more frequently in the combination therapy group (OR 1.45, 95% CI 0.66, 3.19). Overall, 9 patients in each group (9% combination therapy group, 8% monotherapy group) experienced an ADE that led to an antibiotic change or discontinuation (OR 1.10, 95% CI 0.36, 3.32).

Hypothesizing that the 72-hour cutoff for inclusion in the combination therapy group could introduce bias by allowing clinical decline or repeat positive cultures to influence the decision to start combination therapy, a post hoc sensitivity analysis was performed to evaluate a 48-hour cutoff. This analysis included 132 patients in the monotherapy group and 81 in the combination therapy group. Fourteen patients were re-classified from combination therapy to monotherapy with this new cutoff. For the primary composite outcome, this change demonstrated an unweighted OR 2.19 (95% CI 1.22, 3.95) comparing combination therapy to monotherapy, and continued to show no difference between groups when propensity weights were assigned (OR 1.14, 95% CI 0.43, 3.04). The component parts of the primary outcome showed a greater odd of persistent bacteremia (OR 3.96, 95% CI 1.39, 11.22) and no difference between groups for 30-day mortality (OR 0.70, 95% CI 0.23, 2.14) or 30-day recurrence (too few patients to calculate an OR). There was no difference between any secondary outcomes in the propensity score-weighted logistic regression: time to microbiologic cure (mean difference 0.69 days, 95% CI -0.23, 1.88), hospital length of stay (mean difference 4.26 days, 95% CI -8.74, 17.27), and 30-day readmission (OR 1.03, 95% CI 0.46, 2.30).