211 - A Comprehensive Assessment of Carbapenem Use across 90 Veterans Health Administration Hospitals with Defined Stewardship Strategies for Carbapenems

Background: Carbapenems are an important target for antimicrobial stewardship (AS) efforts. In this study, we sought to compare different hospital-based strategies for improving carbapenem use.  

Methods: We analyzed a cohort of all patients hospitalized at Veterans Health Administration (VHA) acute-care hospitals during 2016 and mandatory survey data that characterized each hospital’s carbapenem-specific AS strategy into one of three types: no strategy (NS), prospective audit-and feedback (PAF), or restrictive policies (RP). Sites that could not be classified were excluded. Inpatient carbapenem use was compared across strategies using risk-adjusted generalized estimating equations that accounted for clustering within hospitals. Two Infectious Disease (ID) physicians independently performed manual chart reviews in 425 randomly-selected carbapenem-treated cases (100 for PAF/NS and 225 for RP). Auditors assessed for the presence of ID consultation and carbapenem appropriateness on day 4 of therapy. Assessments were categorized as follows: appropriate (1), acceptable (2), suboptimal (3), unnecessary (4) and inappropriate (5). Assessment scores across strategies were compared with the Kruskal-Wallis test.

Results: There were 429,602 admissions in 90 sites (8 PAF, 24 NS, 58 RP). Median carbapenem use across sites was 17.4 (IQR 8.6-28.4) days of therapy/1,000 days-present. Inpatient carbapenem use was lower at PAF than NS sites [RR 0.67 (95% CI, 0.46-0.98); p=0.04] but similar between RP and NS sites [RR 0.86 (95% CI, 0.61-1.22); p=0.41].
Carbapenem use was considered appropriate or acceptable in 215 (50.6%) of the reviewed cases. Assessment scores were higher (i.e. worse) at NS than RP sites (mean 2.7 vs 2.3; p< 0.01) but did not differ significantly between NS and PAF sites (mean 2.7 vs 2.5; p=0.14).
ID consultations were more common at PAF/RP than NS sites (51% vs 29%; p< 0.01). ID consultations were associated with lower (i.e. better) assessment scores (2.3 vs. 2.6; p< 0.01).

Conclusion: In this VHA cohort, AS strategies and ID consultations were associated with either less or more appropriate carbapenem-prescribing. The use of AS and ID consultations may be complementary, and hospitals could leverage both to optimize carbapenem use.