De-escalation of broad-spectrum therapy has been reported to improve antimicrobial use; however, multi-centered outcomes data associated with de-escalation practice is lacking. Herein, we report 30-day readmission, Clostridium difficile infection (CDI), and length of stay (LOS) associated with antimicrobial de-escalation in patients hospitalized for HCAP.
Data from a retrospective cohort study of patients (n=9319) hospitalized within 119 VA Medical Centers was analyzed (JAC, 2016 Feb;71(2):539-46). Broad-spectrum baseline therapy was defined as hospital day 2 spectrum score > the median (44.5 on a 60 point scale). Propensity score matching was used to pair patients who had broad-spectrum therapy de-escalated with those who continued broad-spectrum therapy beyond hospital day 4. Propensity score models included baseline and clinical covariates associated (P < 0.05 level) with time points prior to de-escalation and outcomes of interest. Logistic regression was used to develop models for each outcome based on significant covariates from the propensity score model and other relevant covariates obtained after day 4 of hospitalization. An indicator variable was added to the model for de-escalation to assess impact. Outcomes are reported as odds ratios (OR) and 95% confidence intervals (CI) or mean (± SD) days.
Propensity matched pairs included 30-day readmission (n=1566), CDI (n=1642), and LOS (n=1575). Cohort demographics included mean (± SD) age of 72.5(12.1) years, 97.8% males, ICU admission rate of 11.6%, and culture-positive rate of 21.7%. Crude 30-day readmission, CDI rates, and mean (± SD) LOS for the cohort (n=9319) were 29.2%, 1.6%, and 7.6 (2.6) days, respectively. After propensity matching and adjustment for covariates antimicrobial de-escalation was not associated with significant difference in 30-day readmission (OR 0.79; 95% CI 0.55, 1.14) or CDI rates (OR 1.44; 95% CI 0.79, 2.63); however, de-escalation was associated with a reduced LOS (Mean difference -0.28 days, 95% CI -0.45, -0.12).
Conclusion: Antibiotic de-escalation was not associated with 30-day readmission or CDI; however, mean LOS was associated with a modest reduction in LOS suggesting that the practice is safe in an HCAP population.
M. Jones, None
M. Samore, None
K. Madaras-Kelly, None
See more of: Poster Abstract Session