Over 23,000 deaths and $20 billion in indirect health-care costs are attributed to resistant bacteria in the United States each year,1 which led to a US Presidential Executive Order in 2014 outlining mitigating actions, one of which was the expansion of antibiotic stewardship programs (ASPs).2
In 2007, the Infectious Diseases Society of America published guidelines outlining 2 core strategies for ASPs; prospective-audit-with-feedback and preauthorization and/or restriction, while recommending investigation to determine the most effective strategy for the pediatric population.3 However, the number of ASPs in pediatric centers is limited4,5 and fewer of these centers have analyzed and published their experience. Most ASPs rely on preauthorization and/or restriction as core strategies6–12 with one focusing solely on prospective-audit and feedback.13
To expand upon previously described strategies, we incorporated 3 distinctive features into our ASP, collectively making the approach unique. These components included: (1) lack of restriction and preauthorization, (2) pharmacist–physician review of all antimicrobials and (3) a rounding-based, in-person approach to feedback carried out by a pharmacist–physician team. Because a handshake provides personal contact and signifies conveyance of trust and sealing of deals, we termed this method “handshake stewardship.” We report evaluation of handshake stewardship on antimicrobial use, both hospital-wide and by unit, in days of therapy per 1000 patient days (DOT/1000 PD), and compare our results to currently described ASPs.
MATERIALS AND METHODS
This study was conducted at Children’s Hospital Colorado (CHCO), a freestanding tertiary care academic pediatric hospital with 444 licensed beds and over 15,000 inpatient admissions yearly. Over 2000 medical staff and 230 medical residents and fellows provide care to patients on the general medical (48 beds), pulmonary and cystic fibrosis (38 beds), hematology/oncology (48 beds), surgical (58 beds), neonatal intensive care (74 beds), pediatric intensive care (32 beds) and cardiac intensive care units (32 beds). Heart, renal, liver and bone marrow transplants (BMTs) are performed.
ASP Design: Handshake Stewardship
The ASP at CHCO began in October 2011 and evolved over 3 phases. The activities and hospital-supported full-time equivalents for members over time are summarized in Figure 1. Stewardship activities were limited in the preimplementation phase, and did not include prospective-audit-with-feedback or restriction. In the second phase (planning phase), a formal program was approved and initiatives broadened to include groundwork for provider acceptance (eg, conduct surveys, meet with section heads, stakeholder recruitment). This phase included guideline development with other departments for common infections. The infectious diseases ASP physician and a pediatric infectious diseases pharmacist completed certificate training in quality improvement.
Phase 3 of the ASP (postimplementation) incorporated the handshake approach with 3 distinct features. First, there are no restrictions and no preauthorization of antimicrobials. Second, all antimicrobials (antibacterials, antifungals and antivirals) administered to inpatients are reviewed by the ASP physician and pharmacist (collectively called “the stewards”). One steward reviews a 24-hour and the other reviews a 72-hour customized antimicrobial report that includes patient information (name, age and weight), antimicrobial information (name, dose, frequency and indication), and the ordering provider’s name and service (daily time commitment: 1 hour/steward). The stewards identify interventions upon review of the report and the patient’s electronic medical record (EMR). Third, the stewards jointly communicate recommendations in-person to providers on the units during clinical rounds. They locate each team and first communicate recommendations between the team’s patient presentation (15 teams total), and then inquire if the team has questions. The stewards locate all teams, even if there are no specific interventions, in order to field questions (daily time commitment: 1–2 hours). Review and rounding occurred 3 days weekly, but increased to 5 days weekly in July 2014.
Antimicrobials were defined as antibacterials, antifungals and antivirals. All antimicrobial data were extracted from the CHCO Infection Control/Epidemiology data warehouse, which houses all patient-level data found in the EMR. Antimicrobial data were expressed as DOT/1000 PD and were aggregated for each month of each phase. Antimicrobials administered by any route, with exception to the topical, ophthalmic, and otic routes, were identified and converted to days of therapy (DOT) based on receiving at least one dose of that antimicrobial on a calendar day (1 DOT).13 The DOT was normalized for patient census using total patient days (PD) as a denominator, and ultimately antimicrobial use was expressed as DOT/1000 PD. For calculation of unit-based use, the unit-specific DOT and PD were used to express a unit-specific DOT/1000 PD.
We performed a retrospective study comparing antimicrobial use during the 3 phases of developing and instituting handshake stewardship. Antimicrobial use was evaluated over the same months within the 3 phases to minimize the influence of seasonal variation on antimicrobial use (preimplementation, October 2010 through September 2011; planning, October 2011 through September 2013; postimplementation, October 2013 through September 2014; Fig. 1). Antibacterial cost (pharmacy expenditures) was reported as total cost per 1000 PD per month using pharmacy invoice data and was obtainable only for the planning and postimplementation phases. To investigate another potential consequence of antimicrobial use, hospital-onset Clostridium difficile rates per 10,000 PD per month14 were collected in the same 2 phases. To confirm that significant decreases in antimicrobial use did not have negative consequences, quality care measures were evaluated for the 3 phases. Using the Pediatric Health Information Systems (PHIS) database for our hospital, length of stay, readmission within 30 days, mortality, All Patient Refined—Diagnosis Related Group severity scores,15 and percent patients ever receiving an antibacterial were collected.
Normality of monthly rates was assessed using histograms. Mean antimicrobial use by month for each of the 3 phases, in DOT/1000 PD, was compared using one-way analysis of variance (ANOVA; primary analysis). Pairwise comparisons were performed using Tukey method. The rate of change in use between the phases was compared using segmented regression (secondary analysis). A time series model confirmed hospital-wide all antimicrobial use trends found by segmented regression. Mean antibacterial cost/1000 PD/month and hospital-onset C. difficile rates/10,000 PD/month were compared between the planning and postimplementation phase by a 2-sample t test. Hypothesis tests were assumed to be 2-sided with a significance level of 0.05. R version 3.1.1 software (R Foundation for Statistical Computing, Vienna, Austria) was used.
Mean Antimicrobial Use: ANOVA
Monthly rates were normally distributed. The mean hospital-wide antimicrobial use decreased by 103 DOT/1000 PD (10.9%) from the preimplementation phase to the postimplementation phase (P < 0.01; Table 1). Antibacterial use decreased by 77 DOT/1000 PD (10.3%; P < 0.01). Both vancomycin and ertapenem use significantly decreased. Meropenem use also significantly decreased, whereas a concomitant increase in other antipseudomonal agents was not detected. Conversely, ceftriaxone use increased by 10 DOT/1000 PD over the study period (P = 0.01).
In comparison with the cardiac and neonatal intensive care units, the pediatric intensive care unit (PICU) had the highest overall use of antimicrobials, but this decreased by 230 DOT/1000 PD (14.5%; P = 0.03; Table 1). PICU vancomycin use also significantly decreased by 83 DOT/1000 PD (28.1%) during the study period (P < 0.01). The hematology/oncology/BMT unit maintained the highest use, but this decreased by 350 DOT/1000 PD (15.9%; P = 0.03). This was largely driven by vancomycin and meropenem use, which decreased by 38.9% and 31.7%, respectively (P < 0.01 and P = 0.03, respectively). Vancomycin use on the general medicine unit decreased by 34.3%, contributing to the 19.3% decrease in antimicrobials on that unit.
Antifungal and antiviral use remained low and did not significantly change during the phases of the ASP, except in the hematology/oncology/BMT unit where antifungal use decreased by 26.8% (874 to 640 DOT/1000 PD; P < 0.01), and antiviral use decreased by 32.9% (353 to 237 DOT/1000 PD; P < 0.01) from the planning phase to the postimplementation phase (data not shown).
Antimicrobial Use Rate of Change: Segmented Regression
Antimicrobial use increased in the preimplementation phase (+4.8 DOT/1000 PD/month), while use decreased in the postimplementation phase (−10.2 DOT/1000 PD/month; P = 0.01; Table 2; Fig. 2). Although preimplementation antibacterial use was declining, antibacterial use declined faster in the postimplementation phase, resulting in a change in slope of −9.9 DOT/1000 PD/month (P < 0.01). The slope of decline of hospital-wide vancomycin use did not significantly vary between phases, despite the significantly lower mean use reported in the ANOVA (Table 2; Fig. 2). Hospital-wide meropenem use increased in the preimplementation phase, but decreased in the planning and postimplementation phases, resulting in a significant change of slope of −3.9 DOT/1000 PD/month. Trends in other antipseudomonal agents were not statistically significant.
Change in slope of total antimicrobial use, antibacterial use, and specific antibacterial use by unit (by segmented regression) are presented in the supplemental table (see Table, Supplemental Digital Content 1, https://links.lww.com/INF/C500).
Antibacterial Cost, Hospital-onset Clostridium difficile and Balancing Measures
Mean antibacterial cost/1000 PD/month from the planning phase to the postimplementation phase did not change significantly (planning: $10,546/1000 PD/month vs. post: $10,451/1000 PD/month; P = 0.93). The mean hospital-onset C. difficile incidence rate/10,000 PD/month decreased significantly from the planning to postimplementation phase (planning: 8.3 rate/10,000 PD to 4.9 rate/10,000 PD; P < 0.01). Throughout the phases, the number of patients receiving antibacterials decreased from 60% to 50%. Mortality rate and length of stay decreased slightly, while severity score and readmission rate increased slightly (Table 3).
In this retrospective study, we report the effectiveness of handshake stewardship, a novel approach designed to extend the value of previously described strategies to decrease antimicrobial use. This strategy includes review of all antimicrobials and incorporates an in-person, rounding-based feedback mechanism by a pharmacist–physician team, but does not include restrictions or preauthorizations.
This approach achieved the lowest reported mean antibacterial use for a pediatric hospital collecting similar antibacterial use data in the literature to date (673 DOT/1000 PD).13 Following implementation, hospital-wide antibacterial use decreased by 10.3%, antifungal use decreased by 12.1%, and antiviral use decreased by 16.4%, for an overall decrease in antimicrobial use of 10.9%. Significant decreases in use were detected on many units.
The widespread decreases achieved after implementation are largely attributable to the handshake stewardship approach rather than unrelated temporal trends or other stewardship-related activities. As use was tracked over 3 phases, it allowed for evaluation of the handshake stewardship approach solely in the postimplementation phase. Decreases in hospital-wide antimicrobial, antibacterial, antifungal and antiviral use did not have significant changes from the preimplementation to planning phases, but had significant decreases from the planning to postimplementation phases. The handshake stewardship approach also resulted in steeper slopes over time, suggesting impact from the approach well beyond what could be considered temporal trends.
Many pediatric hospitals use the antibiotic stewardship report template from PHIS to track and compare their antibacterial use.12 We evaluated our data in this manner and confirmed a decrease; however, it was of a larger magnitude (165 DOT/1000 PD decrease) and to a lower mean (469 DOT/1000 PD) than our results described in this study. This discrepancy is likely explained by differences in data collected; the PHIS report is limited to oral and parenteral administrations of selected antibacterials (aminoglycoside/penicillin, cephalosporin/macrolide, tetracycline/fluoroquinolone and miscellaneous antibiotic/sulfonamide therapeutic categories), whereas our antibacterial data encompasses all use with the exception of antibacterials administered by the topical, ophthalmic and otic routes. Our 26% decrease in all antibacterial use to a mean of 469 DOT/1000 PD is the greatest decrease and the lowest mean use reported when evaluating antibacterial data with the PHIS report.12
There are limitations to this study, including reporting decreased use rather than increased appropriate use. We have attempted to include balancing measures to support the absence of harm due to decreased use, yet in order to do this properly, outcomes are best analyzed by specific diagnoses, which are outside the scope of this study. We are able to report decreases in length of stay and mortality despite an increase in severity scores, which is reassuring; however, it is important to note that variation in population, insurance pressures and seasonal illnesses impact these measures.
A second limitation was an inability to definitively discern the direct effects of handshake stewardship versus activities outside the ASP. For example, C. difficile rates decreased in the postimplementation phase; however, there were rigorous infection control measures implemented concomitantly of paramount importance in this change.
Third is the difficulty in assessing cost savings. We were unable to directly evaluate cost savings as preimplementation antibacterial cost data was unavailable. Antimicrobial use was rising in the preimplementation phase, and it is likely that cost was also increasing, which would have been pertinent to include in our evaluation. We speculate that the lack of significant antibacterial cost savings from planning to postimplementation is largely due to high-cost items. As an example, inhaled tobramycin represented 30% of antibacterial expenditures annually at our hospital, but its use was not significantly impacted by our program, a lesson for future focus. Other antibacterials that were greatly impacted by our program are not as costly (ie, vancomycin and meropenem). Antifungal spend did not decrease from planning to postimplementation (data not shown); however, this is likely multifactorial, including national guidelines recommending more expensive antifungal agents.16,17 Stewardship programs provide cost aversion in many other areas that are difficult to quantify, and it is likely the costs of the program are outweighed by untallied cost savings in terms of fewer hospital acquired infections and drug adverse events.18,19
Another limitation is the lack of intervention data in the postimplementation phase, which would be helpful in comparing process measures. Starting in October 2014, the number of orders and patients reviewed, the number and type of interventions and the acceptance of those interventions were collected. Using that data for context, each month on average the stewards reviewed over 1250 patients and 1600 orders, and recommended nearly 150 interventions with an 84% acceptance rate.
Despite these limitations, our evaluation is comprehensive in nature, with agent and unit-specific data that will be helpful to those looking to expand stewardship activities or to develop an ASP. In our study, the hematology/oncology/BMT and PICU units had high volumes of antimicrobial use and, therefore, may represent important areas of focus as high impact units for ASPs. They had the highest baseline use, but both experienced overall decreases of 14.5% and 15.9%. Notably, decreases in the PICU likely have downstream effects on use in units to which the PICU transfers patients, largely the surgical and ward services. Interestingly, we were unable to impact use on the pulmonary/cystic fibrosis unit, which may be due in part to the presence of national guidelines that dictate appropriate choices and lengths of therapy during a cystic fibrosis exacerbation.20
Our comprehensive evaluation also incorporated assessment of concomitant increases in use of similar agents; for example, a decrease in meropenem could have resulted in a concomitant rise in piperacillin-tazobactam. Our only observed concomitant rise was that of ceftriaxone, which we theorize was likely due to an elective change in treatment guidelines from cefoxitin and/or ertapenem to ceftriaxone plus metronidazole for our 500 appendicitis patients per year. The impact of this guideline in particular is depicted in Table 1, where significant changes for both ertapenem (decrease) and ceftriaxone (increase) were found from the preimplementation to planning phase when the guidelines were instituted. Another strength is our comprehensive statistical analysis using both ANOVA and segmented regression. In our case, it was important to take into consideration both the mean antimicrobial use and the rate of change in use in different phases of implementation. This strength is demonstrated with overall hospital vancomycin use. The differences in mean use were significant (Table 1), whereas rates of change were not (Table 2), likely because use was declining in the preimplementation phase (Fig. 2). We postulate that the declining use in the preimplementation phase correlated with a change from ceftazidime to cefepime for febrile neutropenia in June 2011, as providers were more likely to use concomitant vancomycin due to suboptimal Gram-positive coverage with ceftazidime monotherapy.
Overall, other pediatric hospitals with ASPs have succeeded in decreasing antimicrobial use compared with hospitals without ASPs, though there is variability in efficacy.12 Most programs reported success in using strategies outlined by the national guidelines.6–13 For example, Di Pentima et al6 using both restriction and prospective-audit-with-feedback, described decreases in overall and targeted antimicrobials. Using only restriction, Agwu et al10 described a 14% decrease in antimicrobial use in DOT/1000 PD. Newland et al13 described a 10.9% decrease of all antibacterials (to a mean of 787 DOT/1000 PD) despite only performing prospective-audit-with-feedback on a select group of broad spectrum antibacterials. Although these centers have reported success, it is difficult to directly compare handshake stewardship due to the differences in study methodology. One reported doses administered instead of DOT and aggregated data by year rather than month.6 None described changes in all antimicrobial use (antibacterials, antifungals, antivirals), nor descriptions by unit.6–13 Even with this limited ability to compare, handshake stewardship appears to extend the benefit of the techniques described in these studies.
The distinct features of handshake stewardship were designed to capitalize on direct communication. Anecdotally, this communication resulted in improved relations between infectious diseases and various units, along with creating collaboration for clinical care initiatives and research. The individualized communication engages discussion among providers and promotes education of the medical team, something that prebuilt rules within an EMR cannot do. It is interesting to note that the percentage of patients receiving any antimicrobial decreased by 10% over the study period (Table 3). It is likely an indirect effect of a culture of more judicious prescribing potentially encouraged by the ongoing communication and education provided by the stewards on daily rounds.
We acknowledge the providers and pharmacists at CHCO who welcome stewardship interventions on a daily basis. We also acknowledge the pharmacy, microbiology, epidemiology and infection control departments who all support stewardship initiatives. We acknowledge the transplant infectious disease physicians who have assisted in appropriate and judicious prescribing in the bone marrow, heart, liver and kidney transplant populations. We also thank the IHQSE program for providing invaluable tools in quality improvement design, implementation and research. Amanda L. Hurst assisted in the design of the evaluation, assisted in collection and interpretation of data, drafted the initial manuscript and critically revised the manuscript, approved the final manuscript as submitted and agreed to be accountable for the work. Jason Child and James K. Todd assisted in the design of the evaluation, critically revised the manuscript, approved the final manuscript as submitted and agreed to be accountable for the work. Kelly Pearce and Claire Palmer assisted in the acquisition, analysis and interpretation of the data, critically revised the manuscript, approved the final manuscript as submitted and agreed to be accountable for the work. Sarah K. Parker assisted in the conception and design of the evaluation, assisted in collection and interpretation of data, critically revised the manuscript, approved the final manuscript as submitted and agreed to be accountable for the work.
1. Centers for Disease Control and Prevention. Antibiotic/antimicrobial resistance. Available at: http://www.cdc.gov/drugresistance/
. Accessed May 4, 2015.
2. Obama B. Executive order – combating antibiotic-resistant bacteria. 2014. The White House. Available at: https://www.whitehouse.gov/the-press-office/2014/09/18/executive-order-combating-antibiotic-resistant-bacteria
. Accessed May 4, 2015.
3. Dellit TH, Owens RC, McGowan JE Jr, et al.; Infectious Diseases Society of America; Society for Healthcare Epidemiology of America. Infectious Diseases Society of America and the Society for Healthcare Epidemiology of America guidelines for developing an institutional program to enhance antimicrobial stewardship. Clin Infect Dis. 2007;44:159–177.
4. Hersh AL, Beekmann SE, Polgreen PM, et al. Antimicrobial stewardship programs in pediatrics. Infect Control Hosp Epidemiol. 2009;30:1211–1217.
5. Newland JG, Gerber JS, Weissman SJ, et al. Prevalence and characteristics of antimicrobial stewardship programs at freestanding children’s hospitals in the United States. Infect Control Hosp Epidemiol. 2014;35:265–271.
6. Di Pentima MC, Chan S, Hossain J. Benefits of a pediatric antimicrobial stewardship program at a children’s hospital. Pediatrics. 2011;128:1062–1070.
7. Metjian TA, Prasad PA, Kogon A, et al. Evaluation of an antimicrobial stewardship program at a pediatric teaching hospital. Pediatr Infect Dis J. 2008;27:106–111.
8. Di Pentima MC, Chan S. Impact of antimicrobial stewardship program on vancomycin use in a pediatric teaching hospital. Pediatr Infect Dis J. 2010;29:707–711.
9. Di Pentima MC, Chan S, Eppes SC, et al. Antimicrobial prescription errors in hospitalized children: role of antimicrobial stewardship program in detection and intervention. Clin Pediatr (Phila). 2009;48:505–512.
10. Agwu AL, Lee CK, Jain SK, et al. A World Wide Web-based antimicrobial stewardship program improves efficiency, communication, and user satisfaction and reduces cost in a tertiary care pediatric medical center. Clin Infect Dis. 2008;47:747–753.
11. Sick AC, Lehmann CU, Tamma PD, et al. Sustained savings from a longitudinal cost analysis of an internet-based preapproval antimicrobial stewardship program. Infect Control Hosp Epidemiol. 2013;34:573–580.
12. Hersh AL, De Lurgio SA, Thurm C, et al. Antimicrobial stewardship programs in freestanding children’s hospitals. Pediatrics. 2015;135:33–39.
13. Newland JG, Stach LM, De Lurgio SA, et al. Impact of a Prospective-Audit-With-Feedback Antimicrobial Stewardship Program at a Children’s Hospital. J Pediatric Infect Dis Soc. 2012;1:179–186.
14. Dantes R, Epson EE, Dominguez SR, et al. Investigation of a cluster of Clostridium difficile infections in a pediatric oncology setting. Am J Infect Control. 2016;44:138–145.
15. Iezzoni LI, Ash AS, Shwartz M, et al. Predicting who dies depends on how severity is measured: implications for evaluating patient outcomes. Ann Intern Med. 1995;123:763–770.
16. Pappas PG, Kauffman CA, Andes DR, et al. Clinical Practice Guideline for the Management of Candidiasis: 2016 Update by the Infectious Diseases Society of America. Clin Infect Dis. 2016;62:e1–50.
17. Maertens J, Marchetti O, Herbrecht R, et al.; Third European Conference on Infections in Leukemia. European guidelines for antifungal management in leukemia and hematopoietic stem cell transplant recipients: summary of the ECIL 3–2009 update. Bone Marrow Transplant. 2011;46:709–718.
18. Bates DW, Cullen DJ, Laird N, et al. Incidence of adverse drug events and potential adverse drug events. Implications for prevention. ADE Prevention Study Group. JAMA. 1995;274:29–34.
19. Kaushal R, Bates DW, Landrigan C, et al. Medication errors and adverse drug events in pediatric inpatients. JAMA. 2001;285:2114–2120.
20. Flume PA, Mogayzel PJ Jr, Robinson KA, et al.; Clinical Practice Guidelines for Pulmonary Therapies Committee. Cystic fibrosis pulmonary guidelines: treatment of pulmonary exacerbations. Am J Respir Crit Care Med. 2009;180:802–808.