OBJECTIVE Practice guidelines recommend the shortest duration of antimicrobial therapy appropriate to treat uncomplicated pneumonia be prescribed to reduce the emergence of resistant pathogens. A national evaluation was conducted to assess the duration of therapy for pneumonia. DESIGN Retrospective medication utilization evaluation. SETTING Thirty Veterans Affairs medical centers. PATIENTS Inpatients discharged with a diagnosis of pneumonia. MEASUREMENTS A manual review of electronic medical records of inpatients discharged with uncomplicated community‐acquired pneumonia (CAP) or healthcare‐associated pneumonia (HCAP) was conducted. Appropriate CAP therapy duration was defined as at least 5 days, and up to 3 additional days beginning the first day the patient achieved clinical stability criteria; the appropriate HCAP therapy duration was defined as 8 days. The duration of antimicrobial therapy for intravenous (IV) and oral (PO) inpatient administration, PO therapy dispensed upon discharge, Clostridium difficile infection (CDI), hospital readmission, and death rates were measured. RESULTS Of 3881 pneumonia admissions, 1739 met inclusion criteria (CAP [n = 1195]; HCAP [n = 544]). Overall, 13.9% of patients (CAP [6.9%], HCAP [29.0%]) received therapy duration consistent with guideline recommendations. The median (interquartile range) days of therapy were 4 days (3–6 days), 1 day (0–3 days), and 6 days (4–8 days) for inpatient IV, inpatient PO, and outpatient PO antimicrobials, respectively. CDI was rare but more common in patients who received therapy duration consistent with guidelines. Therapy duration was not associated with the readmission or mortality rate. CONCLUSIONS Antimicrobials were commonly prescribed for a longer duration than guidelines recommend. The majority of excessive therapy was completed upon discharge, identifying the need for strategies to curtail unnecessary use postdischarge. Journal of Hospital Medicine 2015;11:832–839. © 2015 Society of Hospital Medicine

BackgroundComparison of antimicrobial de-escalation rates between healthcare settings is problematic. To objectively and electronically measure de-escalation a method based upon the spectrum of antimicrobial regimens administered (i.e., spectrum score) was developed.MethodsA Delphi process was used to develop applicable concepts. Spectrum scores were created for 27 antimicrobials based upon susceptibility for 19 organisms. National VA susceptibility data was used to estimate microbial spectrum. Susceptibility estimates were converted to ordinal scores, and values for organisms with multi-drug resistance potential were weighted more heavily. Organism scores were summed to create antibiotic-specific spectrum scores and extended mathematically to score multi-antimicrobial regimens. Vignettes were created from antimicrobial regimens administered to 300 patients hospitalized with pneumonia. Daily spectrum scores were calculated for each case. Hospitalization day 4 scores were subtracted from day 2 scores (i.e., spectrum score ∆). A positive spectrum score ∆ defined de-escalation. Experts ranked each pneumonia case on a 7-point Likert scale (Likert >4 indicated de-escalation). Spectrum score ∆s were compared to expert review. Findings were used to identify score deficiencies. Next, 40 pairs of cases were modified to include antimicrobial administration routes. Each pair contained almost similar regimens; however, one contained oral (PO) the other only intravenous (IV) antimicrobials on day 4 of therapy. Experts reviewed cases as described. Spectrum score ∆ credits to account for PO conversion were derived from the mean paired differences in Likert Score. De-escalation status was evaluated in 100 vignettes containing antimicrobial route by different experts and compared to the modified method.ResultsInitial sensitivity and specificity of the spectrum score ∆ to detect expert classified de-escalation events was 86.3 and 96.0%, respectively. In paired cases, the mean (± SD) Likert score was 5.0 (1.5) and 4.6 (1.5) for PO and IV (P = 0.002), respectively. To improve de-escalation event detection, two credits were added to spectrum score ∆s based upon the percentage of antimicrobials administered PO on day 4. The final method, exhibited sensitivity and specificity to detect expert classified de-escalation events of 96.2 and 93.6%, respectively.ConclusionsThe final spectrum score method exhibited excellent agreement with expert judgments of de-escalation events in pneumonia.Electronic supplementary materialThe online version of this article (doi:10.1186/s12879-015-0933-9) contains supplementary material, which is available to authorized users.

De-escalation of antimicrobial therapy was limited and varied substantially across facilities. De-escalation was associated with respiratory tract culture collection and treatment in a high complexity-level facility.

A pharmacy-based triage algorithm helped to reduce excessive DATs for patients with pneumonia at hospital discharge without negatively affecting 30-day readmission rates.