Purpose A protocol to optimize the duration of antimicrobial therapy (DAT) for uncomplicated pneumonia at hospital discharge was evaluated.
Methods This retrospective quasiexperimental study was conducted at Boise Veterans Affairs Medical Center from March 2013 through June 2015. Patients were included in the study if they were diagnosed with pneumonia, were hospitalized for more than 24 hours, received antimicrobial treatment within 48 hours of admission, and survived until hospital discharge. The intervention included development of a pneumonia DAT triage algorithm, a process for assessment of the appropriate DAT by pharmacists, and recommendations to providers to limit excessive discharge DATs prescribed. Interrupted time–series analysis was performed to determine the mean monthly DAT per patient and the 30-day readmission rate.
Results Of the 707 patients discharged with a diagnosis of pneumonia, 560 met the criteria for study inclusion (366 in the preimplementation group and 194 in the postimplementation group). Change in slope of monthly mean DAT per patient postimplementation was significantly reduced (p = 0.03) from the preimplementation slope (p = 0.95), indicating an association between the intervention and mean DAT per patient. The intervention was not associated with the 30-day readmission rate. The mean ± S.D. DAT decreased from 9.5 ± 2.4 days preimplementation to 8.2 ± 2.9 days postimplementation, primarily due to the reduction of outpatient DAT from 5.2 ± 3.0 days preimplementation to 4.2 ± 3.0 days postimplementation.
Conclusion A pharmacy-based triage algorithm helped to reduce excessive DATs for patients with pneumonia at hospital discharge without negatively affecting 30-day readmission rates.
Pharmacists frequently perform medication reconciliation for patients at hospital discharge. Pharmacists take an active role in assisting patients to understand changes in their long-term medication regimens that may be necessary after a hospital stay. While antimicrobials are often included in patients’ discharge medications, medication reconciliation efforts generally do not focus on the total duration of antimicrobial therapy (DAT) that patients receive. Infectious Diseases Society of America (IDSA) guidelines discuss the benefits of limiting DAT for uncomplicated pneumonia, including the minimization of antimicrobial resistance, adverse medication events, and Clostridium difficile infection (CDI).1–3 However, studies have found that the DAT prescribed frequently exceeds IDSA guideline recommendations for shorter courses of therapy.4–6 A medication utilization evaluation (MUE) conducted at our Veterans Affairs (VA) medical center in 2012 indicated that the majority of uncomplicated pneumonia cases were often treated for longer than indicated, with approximately 50% of antimicrobials prescribed during hospital discharge. This MUE was replicated in 30 VA facilities by the National VA Center for Medication Safety (MedSafe) with similar results.6
To remediate the MUE findings, we developed a triage algorithm to allow pharmacists to classify pneumonia complexity and identify an appropriate DAT at discharge. A protocol was developed with instructions for pharmacists performing medication reconciliation at discharge (i.e., discharge pharmacists), and these pharmacists were trained to use the triage algorithm. If the triage algorithm recommendation for the DAT was shorter than the DAT prescribed by the provider, discharge pharmacists were instructed to contact providers and recommend a shortened DAT at discharge. We conducted a study to determine if patient outcomes improved after the implementation of a DAT protocol for all patients discharged with a diagnosis of pneumonia.
This single-center, retrospective quasiexperimental study was conducted at Boise Veterans Affairs Medical Center in Boise, Idaho, from March 2013 through June 2015.
Triage algorithm and note template
The triage algorithm was developed by utilizing available evidence and clinical practice guidelines. The algorithm differentiates three levels of pneumonia complexity: green (uncomplicated), yellow (moderately complicated), and red (very complicated) (Figure 1). The level of complexity was determined by the severity of illness during hospital admission, the development of complications during treatment, the identification of specific pathogens, the concordance between empirical antimicrobial coverage and pathogen susceptibility, and the time required to achieve clinical stability. The algorithm included a recommended DAT range for each complexity level as determined by the discharge pharmacist. An electronic note template was developed for the pharmacist to document clinical decision-making regarding evaluation of the pneumonia DAT and to capture documentation of recommendations made to the provider if necessary (Figure 2).
An excessive duration of antimicrobial therapy for pneumonia is frequently prescribed at hospital discharge for regimen completion in the outpatient setting.
Use of a pharmacist-driven intervention with antimicrobial stewardship support can decrease unnecessary antimicrobial use without altering readmission rates.
Discharge medication reconciliation may be a novel target for antimicrobial stewardship programs to reduce unnecessary antimicrobial use.
Medication reconciliation process
At our facility, discharge orders are submitted through the computerized patient record system (CPRS) by the provider. Discharge medication orders are automatically printed in the pharmacy, and the discharge pharmacist is alerted to the impending discharge. The pharmacist performs medication reconciliation by reviewing the electronic medical record for possible medication errors or discrepancies and contacting the prescriber to clarify or correct questionable orders, if necessary. Next, medication reconciliation is documented in the CPRS using the electronic discharge note template. Finally, the discharge medication orders are processed, and a bedside consultation is provided to the patient addressing new, changed, or discontinued medications.
The pneumonia protocol differed from usual care by requiring the discharge pharmacist to identify pending discharges for patients being treated specifically for pneumonia. Pharmacists were instructed to add the days of inpatient i.v. and oral antimicrobial therapy administered and outpatient oral or i.v. antimicrobial therapy prescribed in the discharge orders, which provided the total DAT the patient was to receive. Discharge pharmacists were instructed to use the DAT algorithm to classify the severity of pneumonia and identify the appropriate DAT. If the prescribed DAT exceeded the DAT indicated by the algorithm, pharmacists were instructed to contact the prescriber and recommend a DAT concordant with the algorithm. For example, if a patient with uncomplicated community-acquired pneumonia (CAP) received 5 days of i.v. and subsequent oral antimicrobial therapy as an inpatient and was prescribed an additional 7 days of oral therapy at discharge (total DAT of 12 days), the pharmacist would call the provider and recommend changing the order to 2 days of oral outpatient therapy (for a total DAT of 7 days). If the provider was not available, discharge pharmacists were instructed to wait up to one hour for a callback before filling the prescription as written. The pharmacist then completed the CPRS discharge note using the template designed specifically for pneumonia discharges and conducted the bedside patient consultation per usual care. Pharmacists were instructed to contact the antimicrobial stewardship pharmacist if they were uncertain about how to assess the complexity of specific pneumonia cases.
All discharge pharmacists completed a one-time live training session on the use of the protocol, the triage algorithm, and the note template. During the first month after implementation, the antimicrobial stewardship pharmacist also reviewed and discussed the cases with the discharge pharmacists. In addition, a one-hour inservice education program on the pneumonia discharge process was conducted, and copies of the pneumonia triage algorithm were provided to the medical resident staff.
Patient eligibility and data collection
The electronic medical records of all inpatients with a discharge diagnosis of pneumonia as determined by International Classification of Diseases, Ninth Revision, Clinical Modification codes 480–486 admitted between March 2013 and June 2015 were reviewed. Patients were included in the study if they were hospitalized for more than 24 hours, received antimicrobial treatment within 48 hours of admission, and survived until hospital discharge. Patients were excluded from the analysis of the protocol if they were directly transferred after hospitalization to an outside hospital, completed their hospitalization at an outside hospital (e.g., Fee-Basis admission), were readmitted within 24 hours after discharge, had a very complicated case of pneumonia (red classification), were hospitalized for more than 14 days, or transitioned to hospice or died during hospitalization. In addition, patients for whom the total DAT prescribed could not be determined through chart abstraction or no clinical evidence of pneumonia for the current admission was found were also excluded from the analysis. Very complicated pneumonia was defined as severe immunosuppression (transplant patients, HIV infection, acquired or congenital immunodeficiency, absolute neutrophil count of <500 cells/mm3), empyema, lung abscess, or parapneumonic effusion.
Patient data collected included demographics, prior healthcare exposures, comorbidities, vital signs and laboratory test values indicating clinical stability, positive culture and susceptibility results, days of inpatient i.v. and oral antimicrobial therapy administered, outpatient antimicrobials dispensed at discharge or recommended for outside fill at discharge (for patients not receiving medication from our facility), and readmission within 30 days of DAT completion.
The primary endpoint was the mean monthly DAT per patient; the primary safety endpoint was the 30-day readmission rate. These endpoints were measured as monthly time–series. Secondary outcomes included the mean duration of inpatient i.v. and oral antimicrobial therapy and the mean duration of oral antimicrobial therapy prescribed at discharge before and after protocol implementation. Additional outcomes included an assessment of the time needed by the discharge pharmacist to complete the pneumonia medication reconciliation and the discharge pharmacist’s opinion of operational perspectives (e.g., ease of protocol use, workflow integration).
A time–motion assessment was used to evaluate the time needed by discharge pharmacists to perform medication reconciliation for a sample of pneumonia patients before and after implementation of the protocol. The difference in the time between signed provider discharge orders and signed pharmacy discharge note was calculated. The difference between these times was expressed as a mean time per discharge for the pharmacist to complete the intervention.
Discharge pharmacists’ opinions on the process
A voluntary focus group session was facilitated by a noninvestigator or nondischarge pharmacist approximately five months after protocol implementation to gain feedback from discharge pharmacists regarding the process and ease of use of the protocol. The investigators created questions and discussion points for the focus group but were not present during the session. Responses were transcribed and compiled into common themes by the facilitator in order to ensure pharmacist anonymity.
Interrupted time–series (ITS) analysis was performed for the primary endpoints, comparing the 18-month preimplementation period to the 9-month postimplementation period. ITS analysis was used to estimate trend in an endpoint preimplementation, change in trend postimplementation, and change in endpoint level coinciding with the first month of the intervention. The mean monthly DAT time–series was modeled with linear regression. Thirty-day readmission counts per month were modeled in the ITS with Poisson regression with an offset to account for differences in numbers of patients across months. Demographics and secondary endpoints were characterized using descriptive statistics. Analyses were conducted within the Veterans Informatics Computing Infrastructure utilizing R, version 3.1.2 (R Development Core Team, Vienna, Austria) and SPSS Statistics, version 22.214.171.124 (IBM Corporation, Armonk, NY).
This research complied with all federal guidelines and Department of Veterans Affairs policies relative to human subjects research.
During the 27-month study period, there were 707 patients with a discharge diagnosis of pneumonia. Of those, 560 patients met inclusion criteria for analysis: 366 in the preimplementation period and 194 in the postimplementation period. The exclusion rates pre- and postimplementation were 21.5% and 19.5%, respectively (p = 0.36). Primary reasons for exclusion from the analysis included no evidence of pneumonia during the current hospital visit or admission for pneumonia in another hospital (11.7%), transition to hospice or death during hospitalization (4.2%), complicated pneumonia due to immunosuppression or structural lung changes in the lungs such as empyema or abscess (3.4%), inability to determine the total DAT (2.1%), and admission for >14 days (1.0%). Some patients met multiple exclusion criteria.
Patients were generally elderly and male and had similar characteristics in the pre- and postimplementation phases of study (Table 1). Patients with prior healthcare exposures (e.g., healthcare-associated pneumonia) were numerically more common in the preimplementation phase (p = 0.07); however, the distribution of uncomplicated and moderately complicated cases between study periods was similar. Primary reasons for designation of moderately complicated (yellow) cases included culture of gram-negative nonenteric bacteria from admission cultures (33%) and failure to reach clinical stability criteria by day 4 (25%).
Before implementation of the discharge protocol, the monthly mean DAT per patient was stable (slope preimplementation, ≤0.01; 95% confidence interval [CI], −0.05, 0.05) (Figure 3). After initiation of the discharge protocol, there was a negative change in slope (Δ slope postimplementation, −0.20; 95% CI, −0.36, −0.03), indicating that the implementation of the protocol was associated with a reduction in the monthly mean DAT per patient. For the primary safety endpoint, the 30-day readmission rate before the intervention was stable (slope preimplementation <−0.01; 95% CI, −0.04, 0.04) (Figure 4). The change in slope postimplementation was not statistically significant (Δ slope postimplementation, −0.04; 95% CI, −0.48, 0.63), providing no evidence that the intervention was associated with a change in 30-day readmission rate. The mean ± S.D. total DAT before implementation of the DAT protocol was 9.5 ± 2.4 days, which decreased to 8.2 ± 2.9 days postimplementation. Similarly, the median (interquartile range [IQR]) DAT was reduced from 9 days (7–11 days) to 7 days (6–9 days) postimplementation. Uncomplicated (green) cases generally had a shorter duration than somewhat complicated cases (yellow). Many of the patients with Pseudomonas aeruginosa were treated for at least 14 days, and the percent reduction in the DAT was similar regardless of pneumonia severity classification (Table 2). While the inpatient DAT was similar before and after implementation, the mean ± S.D. outpatient DAT decreased from 5.2 ± 3.0 days preimplementation to 4.2 ± 3.0 days postimplementation. The reduction in the DAT was most notable for CAP outpatient therapy. The pneumonia-specific electronic note template was completed by the discharge pharmacist in 130 patients (67%) discharged with a pneumonia diagnosis. Of cases without the documented pneumonia discharge notes, 20 (31%) had received their entire DAT during hospitalization. The DAT indicated in the discharge orders was longer than recommended by discharge pharmacists using the pneumonia discharge algorithm in 42 cases (32%). Providers accepted recommendations made by discharge pharmacists to shorten the DAT in 19 cases (45%), including changing the algorithm-recommended DAT in 14 cases (33%) and shortening the DAT to shorter than initially prescribed but greater than the algorithm-recommended duration in 5 cases (12%). Pneumonia complexity classifications made by discharge pharmacists as documented in the note template were discordant with investigator chart review in 14 cases (11%). In 10 of these cases, the pharmacist made a more complex pneumonia determination than did investigator chart review utilizing the triage algorithm.
A total of 26 patient discharges were audited (13 preimplementation and 13 postimplementation) during the time–motion study. Discharge times ranged from 0.9 to 4.7 hours and from 1.2 to 3.7 hours for the preimplementation and postimplementation groups, respectively. The mean ± S.D. times per discharge were 2.8 ± 1.4 hours for the preimplementation group and 2.7 ± 0.7 hours for the postimplementation group (p = 0.76).
The focus group comprised six discharge pharmacists who identified possible improvements to the process while expressing positive experiences with the algorithm and template (Table 3). Pharmacists felt that providers generally reacted favorably when recommendations were made. Also, the process was believed to add about two minutes to the time it took to complete a discharge, with green cases being the quickest to complete. Potential areas for improvement included increasing accessibility to antimicrobial stewardship personnel (e.g., weekends) and moving the note template to a more accessible location in the electronic medical record. Using this feedback, the investigators made some modifications to the location of the note template and clarified some portions of the algorithm decision-making template to improve ease of use.
This study found that staff pharmacists performing discharge medication reconciliation for patients with pneumonia could reduce an excessive DAT without affecting the readmission rate. The reduction in the DAT occurred primarily through a decrease in therapy prescribed upon discharge from the hospital. The findings are significant for several reasons. First, few patients complete their entire course of antimicrobial therapy in the hospital, and approximately 50% of therapy for pneumonia is prescribed at discharge. The aforementioned VA National MUE identified mean ± S.D. DATs of 4.2 ± 2.5, 1.6 ± 2.0, and 6.0 ± 4.2 days for inpatient i.v., inpatient oral, and outpatient oral antimicrobials, respectively, prescribed at discharge.6 Further, antimicrobial stewardship program personnel frequently contact providers with recommendations for therapy deescalation or i.v.-to-oral conversion during the change of therapy rather than at the completion of therapy. Many uncomplicated cases may not require intervention by antimicrobial stewardship personnel during the change period; therefore, antimicrobial stewardship personnel may not be aware that an excessive DAT is being prescribed. Medication reconciliation during transitions of care is routinely performed by staff pharmacists, and recently the Centers for Medicare and Medicaid Services proposed rules for skilled nursing facilities that require pharmacists to review records of patients receiving antimicrobials during transitions of care from other healthcare facilities.7 Our analysis suggests that with appropriate guidance from antimicrobial stewardship personnel, staff pharmacists may accomplish this by intervening at the completion of therapy to reduce excessive DATs. Second, it is important to recognize that most CDI is acquired after healthcare exposure, and an excessive DAT is strongly linked to the development of CDI.8–10 Reducing excessive DATs at hospital discharge may help reduce the frequency of CDI; however, our study was not powered to evaluate this outcome, as there were fewer than 12 cases of healthcare-acquired CDI in the entire facility during the study period.
One strength of our study was the development of an evidence-based triage algorithm that provided a framework for pharmacists to evaluate patient complexity and clinical improvement when determining the appropriate DAT. The algorithm was developed by the investigators for use in this study but was subsequently refined with input from the VA Antimicrobial Stewardship Task Force.11 The pneumonia triage algorithm, the note template, and a clinical guidance document were made available to all VA facilities wishing to adopt a similar protocol through a national VA webinar and SharePoint site. Other strengths include the incorporation of the protocol into the existing medication reconciliation process in use by staff pharmacists and the methods of analysis. The medication reconciliation process already required pharmacist clinical decision-making for chronic conditions and contacting of providers to clarify orders. The discharge pharmacists were able to assess uncomplicated cases quickly, adopting the process into their workflow. ITS is a preferred method of analysis for quasiexperimental study design of facility-level interventions.12 When combined with the aggregated changes in the DAT, process evaluations provided a detailed description of the study findings.
Study limitations included the single-center design in a small VA setting, a modest acceptance rate of triage algorithm-based recommendations, and inconsistent use of the pneumonia-specific note template for documentation. In general, the bed size of VA facilities is smaller than non-VA facilities, and VA centers dispense outpatient medications to the majority of patients during hospital discharge. Further, in the aforementioned national MedSAFE pneumonia MUE, 80% of the participating VA centers used pharmacists to perform discharge medication reconciliation. In addition, the appropriateness of antibiotic selection at discharge was not evaluated. However, antibiotic appropriateness is reviewed Monday through Friday by antimicrobial stewardship pharmacists during hospitalization, and discharge pharmacists were encouraged to contact the the pharmacist if they had questions about agent selection or DAT. Caution is advised when generalizing the findings to dissimilar settings, but preliminary findings using a somewhat similar approach in a 500-bed community hospital revealed similar results.13 While the acceptance rate of pneumonia algorithm–based DAT recommendations was relatively low, providers did compromise and reduce the DAT in additional cases. Further, the providers received education about the DAT for pneumonia and the pneumonia triage tool, and they were informed that pharmacists might contact them with recommendations to reduce the DAT. The monthly mean DAT per patient decreased steadily during the intervention months, which suggests that both pharmacists and clinicians gained confidence with the protocol. Finally, pneumonia-specific discharge templates were not completed without a specific reason in approximately 20% of eligible cases. Findings from the focus group suggested that the pharmacist simply forgot and utilized the regular discharge template in some of these cases. In other cases there was confusion about the need to use the template for patients discharged who did not receive medications at discharge from the VA facility.
There are few examples of interventions designed to limit excessive durations of therapy in the literature, particularly for antimicrobials prescribed at hospital discharge. Patient-level prospective audit with feedback may be effective in limiting pneumonia DATs. Avdic et al.4 evaluated the management of presumed CAP before and after education and prospective feedback to medical teams concerning antimicrobial selection and duration of therapy. This study defined the appropriate DATs as 5 days for patients without immunocompromise or structural lung diseases, 7 days for patients with moderate immunocompromise or structural lung diseases, and 10–14 days for patients who had a poor clinical response, received initial inappropriate therapy, or were significantly immunocompromised. The intervention led to a decrease in the median (IQR) duration of therapy from 10 days (8–13 days) to 7 days (7–8 days) without increasing clinical failure or readmission rates, comparable to our findings.
Serial procalcitonin measurements have also been used to guide the DAT for pneumonia. A Cochrane analysis of 14 studies of patients with acute respiratory infections found that procalcitonin-guided therapy compared with usual care resulted in a reduction in the median total DAT from 8 to 4 days.14 In contrast, a multicenter international study found that a procalcitonin-guided algorithm to direct therapy was associated with a reduction in the rate of antimicrobial use for respiratory tract infections; however, predictors of the total duration of antibiotic therapy included compliance with the protocol, which was less than 40% in U.S. centers.15 The cost:benefit ratio of procalcitonin measurement is unclear.16 Procalcitonin was not used to guide the DAT in our study.
Future research should include assessment of the protocol in other practice settings in which pharmacists perform discharge medication reconciliation. Assessments should include sufficient sample size to provide a better estimate of protocol safety, including the impact on CDI rates. Further, development of similar algorithms for common infectious discharge diagnoses, such as uncomplicated urinary tract infections and skin and soft tissue infections, is warranted.
A pharmacy-based triage algorithm helped to reduce excessive DATs for patients with pneumonia at hospital discharge without negatively affecting 30-day readmission rates.
The Veterans Affairs Antimicrobial Stewardship Task Force is acknowledged for review of the pneumonia triage algorithm and related materials. The discharge pharmacists at the Boise Veterans Affairs Medical Center are acknowledged for participating in the intervention.
Dr. Caplinger, Ms. Wilkin, and Mr. Remington received salary support from the ASHP Research and Education Foundation Federal New Investigator Award Program paid through Idaho State University for this work. Dr. Madaras-Kelly is employed by Idaho State University, holds a “without compensation appointment” at the Boise Veterans Affairs Medical Center, and receives grant support unrelated to this work from the National Institutes of Health and the Department of Veterans Affairs. The authors have declared no other potential conflicts of interest.
The views expressed in this article are solely those of the authors and do not necessarily reflect the position or policy of the Department of Veterans Affairs or ASHP.
- Copyright © 2016 by the American Society of Health-System Pharmacists, Inc. All rights reserved.