OBJECTIVE: To prospectively evaluate the effect of a comprehensive antimicrobial stewardship program on antimicrobial use, physician interventions, patient outcomes, and rates of antimicrobial resistance.
METHODS: Active surveillance of antimicrobial use with intervention and real-time feedback to providers and reinforcement of prior authorization for selected antimicrobials were introduced at a pediatric teaching hospital. Antimicrobial-use indications were incorporated as a mandatory field in the computerized information system. An automated report of antimicrobials prescribed, doses, patient demographics, and microbiology data was generated and reviewed by an infectious-disease pharmacist and a pediatric infectious-disease physician. Antimicrobial use, expressed as the number of doses administered per 1000 patient-days, was measured 3 years before and after the implementation of the program.
RESULTS: Total antimicrobial use peaked at 3089 doses administered per 1000 patient-days per year in 2003–2004 before implementation of the program and steadily decreased to 1904 doses administered per 1000 patient-days per year during the postintervention period. Targeted-antimicrobial use declined from 1250 to 988 doses administered per 1000 patient-days per year. Nontargeted-antimicrobial use declined from 1839 to 916 doses administered per 1000 patient-days per year. Rates of antimicrobial resistance to broad-spectrum antimicrobials among the most common Gram-negative bacilli remained low and stable over time.
CONCLUSIONS: The successful implementation of antimicrobial stewardship strategies had a significant impact on reducing targeted- and nontargeted-antimicrobial use, improving quality of care of hospitalized children and preventing emergence of resistance.
WHAT'S KNOWN ON THIS SUBJECT:
Few children's hospitals have implemented comprehensive antimicrobial stewardship programs (ASPs). Beyond reporting proportions of children who received antimicrobials during their hospitalization, complete data about the extent of antimicrobial use in pediatric hospitals before and after ASP implementation do not currently exist.
WHAT THIS STUDY ADDS:
Since introducing an ASP in 2004, its effect on antimicrobial use, interventions, outcomes, and rates of antimicrobial resistance has been prospectively evaluated. This publication thoroughly describes the impact of an ASP in a children's hospital over a 3-year period.
Rising concerns about antimicrobial resistance and inadequate development of effective new anti-infective drugs have stimulated universal efforts to strengthen infection-control interventions and antimicrobial stewardship practices.1,–,4 Since the introduction of antimicrobial stewardship programs (ASPs) in US hospitals in the 1980s, these programs have been pivotal in reducing unnecessary antimicrobial use.5,–,10 In 2007, the Infectious Diseases Society of America, in collaboration with other professional organizations, released guidelines to assist in the implementation of multidisciplinary ASPs.7 However, despite the overwhelming evidence supporting antimicrobial management practices in adult health care centers, few children's hospitals have implemented comprehensive programs, and data about the effect of ASPs in pediatric settings have been limited.11,–,17 Furthermore, beyond reporting proportions of hospitalized children who have received antimicrobials during their hospitalization, complete data about the extent of antimicrobial use in pediatric hospitals before and after ASP implementation do not currently exist.18
Since introducing an ASP in 2004, we have prospectively evaluated its effect on antimicrobial use, physician interventions, patient outcomes, and rates of antimicrobial resistance. To the best of our knowledge, this is the first publication to thoroughly describe the impact of an ASP in a children's hospital.
Antimicrobial Stewardship Program
In March 2004, the Medical Executive Committee of the Alfred I. duPont Hospital for Children approved the implementation of an inpatient ASP beginning April 1, 2004. The ASP team included a full-time doctoral-level clinical pharmacist with postdoctoral training in infectious diseases and a physician director who was a board-certified pediatric infectious-disease practitioner.11,12
During the first year of the program, 16 antimicrobials were targeted for prospective audit on the basis of their spectrum of activity and cost (Table 1). After the first year of the ASP, all prescribed antimicrobials were monitored prospectively. The most commonly used narrow-spectrum or less-expensive antibiotics and antifungal agents were termed “nontargeted antimicrobials” and were also monitored prospectively (Table 1).
Before implementation of the ASP, cefepime and meropenem required prior authorization by the attending pediatric infectious-disease physician, but antimicrobial use and compliance with prior authorization were not monitored. After implementation of the ASP, prior authorization by the attending infectious-disease physician was extended to also include linezolid and voriconazole, which were added to the hospital formulary at that time. The name of the approving physician was required as a mandatory field in the computerized prescriber order entry (CPOE) system and retrieved daily by the infectious-diseases pharmacist and confirmed with the approving physician to ensure compliance.
As part of the ASP, the formulary was revised, and piperacillin-tazobactam was added to replace ticarcillin-clavulanic acid. Ceftazidime was removed from the formulary, and the prior-authorization requirement for cefepime was eliminated.
Antimicrobial use is expressed as the total number of individual antibiotic or antifungal doses administered per year. Data from 3 years before the implementation of the program, April 1, 2001, to March 31, 2004, were used to determine the trend of antimicrobial use during the preintervention phase and to compare with the postintervention period from April 1, 2004, to March 31, 2007.
The study was conducted at the Alfred I. duPont Hospital for Children, a 180-bed tertiary care academic pediatric hospital affiliated with Thomas Jefferson University (Philadelphia, PA). The inpatient ICUs, which comprised 26% of the total hospital beds, included a 15-bed NICU, a 21-bed PICU, and a 10-bed cardiac ICU. Oncology patients were housed in a 22-bed unit, and patients who were receiving stem cell transplantations were housed in a 10-bed stem cell transplantation unit, both of which comprised 18% of the total hospital beds.11 Overall, the inpatient units averaged 9000 admissions per year; 78 pediatric/medicine residents and 31 pediatric fellows provided rotating care. Stationary and mobile computer terminals for accessing the computerized information system were widely available. A CPOE system (CareNet [Cerner Corporation, Kansas City, MO]) was introduced in 1999.11,12
Antimicrobial guidelines, including indications and dosages for pediatric patients and renal adjustments, were developed by the ASP team and made available as part of the hospital formulary.11,12 A printed pocket version was distributed among hospital staff and residents 1 month before starting the ASP.11,12 Approved indications for all antimicrobials were included as a mandatory field in the CPOE. Selection of an antimicrobial indication was required before proceeding with the ordering process.
Case Findings and Review
An automated report of all hospitalized patients who had received antimicrobials within the previous 24 hours was generated daily, except on weekends. A prospective audit of antimicrobials prescribed during the weekend was performed on Mondays. Critical data elements extracted from the informatics system included type of antimicrobial; dose, interval, and route of administration; patient age, weight, allergies, and renal function; attending physician and admitting service names; and location within the hospital. Microbiology data, including all positive culture results from any site, were reported daily for all hospitalized patients. In addition, annual antibiograms were generated from the automated Misys system (Misys Healthcare Systems, Inc, Tucson, AZ). The infectious-disease pharmacist reviewed all of these reports and the medical records when indicated and discussed these data with the ASP medical director. Feedback, including alternative antimicrobial therapy and dosing recommendations, was performed through direct 1-on-1 communication with the infectious-disease pharmacist and the resident, fellow, or attending physician who prescribed the antibiotics. The prescriber was allowed 24 hours to implement the suggested modifications. If the recommendations were not implemented at the end of the 24-hour time period, a second telephone call was made by the medical director of the ASP directly to the attending physician. For cases in which the infectious-disease service was consulted, recommendations were discussed with the attending infectious-disease clinician.
ASP recommendations were based on standardized evidence-based practices defined by national guidelines and various subspecialty practices and were recorded as interventions. Interventions that qualified as prescribing errors were reported in a standardized fashion, as described previously.11,12
Measures of Antimicrobial Use
Medication administration record data were stored in the Cerner database and in the Nemours data warehouse. Tables containing selected critical data elements were downloaded daily from CareNet and PharmNet to the Nemours data warehouse by using an Oracle database (Oracle Corporation, Redwood Shores, CA) established to integrate business, operational risk, and clinical data with patient encounters. Numbers of antimicrobial doses administered were retrieved by querying the Cerner medication administration record tables linked to the antimicrobial generic names in the data warehouse. Data were captured by number of doses administered to each unique patient. Doses administered were normalized per 1000 patient-days to control for differences in the annual hospital census.
Patient data, antimicrobial interventions, and outcomes were entered into a spreadsheet database (Microsoft Excel 2000 [Microsoft, Redmond, WA]). Annual antimicrobial use was calculated as doses administered per 1000 patient-days. Acuity of patient care was determined by the number of pediatric intensive care admissions per 1000 hospital admissions per year. Segmented regression analysis was performed to compare the temporal change for preintervention and postintervention antimicrobial use. In addition, a χ2 trend test for proportion was used to test the trend in antimicrobial use applying 2003–2004 data as baseline. All tests were 2-tailed with a .05 level of significance. Analyses were performed by using statistical software SPSS 17.0 (SPSS Inc, Chicago, IL) and R 2.8.1.
Over the 6 years of the study, 3 years before and 3 years after the ASP, a total of 482 377 doses of antimicrobials were administered to 27 214 unique children (48% of all hospitalized children). During the preintervention period, 44% (11 180 of 25 680 hospital admissions) of hospitalized children received antimicrobials; the medians were 4 doses of targeted (range: 1–366; interquartile range [IQR]: 2–9) and nontargeted (range: 1–191; IQR: 2–8) antimicrobials administered per patient per hospitalization. After implementation of the ASP, 43% (13 100 of 30 862 hospital admissions) of children received antimicrobials, but the median number of doses remained unchanged at 4 doses of targeted (range: 1–172; IQR: 2–9) and nontargeted (range: 1–193; IQR: 2–8) antimicrobials administered per patient per hospitalization.
Figure 1 shows the trends of antimicrobial use 3 years before and 3 years after ASP implementation. Total antimicrobial use peaked in 2003–2004 before ASP implementation and steadily decreased over the next 3 years of the program (Table 2). Targeted-antimicrobial use increased during the 3 years before the implementation of the ASP from 557 to 1250 doses administered per 1000 patient-days per year and declined during the postintervention period by 21%, to 988 doses administered per 1000 patient-days per year (P < .001). Among targeted antimicrobials, a significantly greater decline in antibiotic use was noted for antimicrobials that required prior authorization (36%) in comparison to the decline in antimicrobials prescribed without prior authorization (19%). Table 2 shows the trends of targeted-antimicrobial use according to class after implementation of the ASP.
During the preintervention period, use of nontargeted antimicrobials increased from 753 to 1839 doses administered per 1000 patient-days per year. Table 2 depicts the trend of nontargeted-antimicrobial use during the postintervention years. Nontargeted-antibiotic use peaked at 1788 doses administered per 1000 patient-days per year before implementation of the ASP and decreased to 912 doses administered per 1000 patient-days per year by 2007 (χ2 trend: 467; P < .0001). Among these nontargeted antibiotics, the use of gentamicin and penicillins (oxacillin, ampicillin, and ampicillin-sulbactam) decreased from 372 and 683 doses administered per 1000 patient-days per year before the introduction of the ASP to 92 and 255 doses administered per 1000 patient-days per year during the last year of the study, respectively (P < .001).
The use of nontargeted antifungal agents (itraconazole and amphotericin B deoxycholate) peaked at 50 doses administered per 1000 patient-days per year before the ASP and decreased to 4 doses administered per 1000 patient-days per year (χ2 trend: 82; P < .0001) after introduction of the ASP.
Interventions and Outcomes
The ASP team initiated 1673 interventions in 973 unique patients from 2004 to 2007 (3% of all hospital admissions). The mean age of the children with interventions was 7.2 years (median: 5 years; range: 0 days to 22 years). The male-to-female ratio was 1.1. Dose adjustment (40.2% [672 of 1673]) and modification of antimicrobial therapy (34.8% [583 of 1673]) were the most common interventions performed, followed by parenteral-to-oral antibiotic conversion (10% [167 of 1673]), additional laboratory monitoring (6.5% [108 of 1673]), infectious-disease consultation (5.8% [97 of 1673]), and consultation regarding pharmacokinetics and reinforcement of the prior-authorization policies (2.7% [46 of 1673]). Among the dose-adjustment interventions, dosage and frequency modification represented the majority of interventions (96.1% [646 of 672]). The remaining interventions were renal dose adjustments (3.9% [26 of 672]).
No significant differences were noted in the number or category of interventions performed by the antimicrobial stewardship team over time, with the exception of parenteral-to-oral interventions, which represented 23% of all interventions performed during the first year of the ASP but only 0.9% of interventions during the last year of the program (χ2 trend: 5.97; P = .015).
Modification of antimicrobial therapy involved recommendation of an alternative antimicrobial therapy (43.1% [251 of 583 interventions]), followed by discontinuation of antimicrobial therapy (34.8% [203 of 583]), prolonged duration of therapy (11.3% [66 of 583]), and adding antimicrobial therapy (10.8% [63 of 583]). No significant differences were noted over time.
The most frequent antimicrobials that required interventions included vancomycin (16% [268 of 1673]), piperacillin-tazobactam (11% [184 of 1673]), cefepime (6.8% [114 of 1673]), ceftriaxone (6.3% [105 of 1673]), and fluconazole (5% [82 of 1673]).
Outcomes associated with ASP interventions are listed in Table 3. Over time, the rate of compliance with the ASP recommendations increased from 83% to 92% (χ2: 24.22; P < .001).
From 2003 to 2009, the rates of antimicrobial resistance to broad-spectrum antimicrobials among the most common Gram-negative bacilli did not change significantly (Fig 2, Table 4). Enterobacter cloacae susceptibility to piperacillin-tazobactam in 2003, before the introduction of this antibiotic to our hospital, was 71%. In 2009, E cloacae susceptibility to piperacillin-tazobactam was 81%.
Despite efforts to control antimicrobial use, the number of antibiotic prescriptions made for hospitalized adults continues to increase.19 A similar result was found recently in a study of hospitalized pediatric patients in which linezolid and macrolide use significantly increased over time; there were no reductions in other antibiotic use from 2002 to 2007.18 In our experience, implementation of a comprehensive ASP led to significant reductions of both targeted and nontargeted antimicrobials and modifications of the patterns of antimicrobial use after ASP implementation. Total antimicrobial use decreased by 38% despite a 7% increase in the acuity of patient care from 2004 to 2007. Furthermore, targeted antimicrobials that required prior authorization (36%) had a more significant decline when compared with nonrestricted antimicrobials (19%).
Less change was noted in the rates of targeted antifungals after the ASP was initiated, because there has been decreasing use of liposomal amphotericin B formulations globally, which was also experienced in our hospital.20,–,22 With the addition of voriconazole to our hospital formulary, prior authorization by the infectious-diseases attending physician was required, which is why there was not a significant increase in azole use during the postintervention period. Echinocandins and posaconazole were not prescribed during the initial 3 years of the ASP. Moreover, the only antifungals considered nontargeted by the ASP were itraconazole and amphotericin B deoxycholate, which are seldom used in hospitalized children, as is reflected in the significant decline in the use of these agents.
We were able to achieve a significant reduction for all classes of targeted antibiotics except for quinolones, which remained low and did change significantly over the course of this study.23 During the postintervention period, use of vancomycin decreased by 54%, whereas the use of linezolid remained stable.11,24
Over time, the rates of noncompliance with antimicrobial stewardship recommendations decreased by >50%. The 92% rate of compliance with antimicrobial recommendations was higher than previously reported by other investigators.25,26 The 1-on-1 dialogue between members of the ASP team and the medical staff might have affected these high rates of compliance over the successive years. In addition, direct communication provides an opportunity for education to reduce unnecessary antimicrobial use.27 The fact that there was a significant decrease in parenteral-to-oral interventions, and the spontaneous change to oral antibiotics occurred more frequently after the first year of the program, indicates that behavior was being modified.
Evaluating the impact of the ASP on institutional antimicrobial resistance has been challenging.6,7 During the years reported, the antibiotic susceptibilities to broad-spectrum antibiotics for the most common Gram-negative pathogens isolated in our hospital remained high and stable. Moreover, the rates of methicillin-resistant Staphylococcus aureus and vancomycin-resistant enterococci bloodstream infection decreased.24,28,29 In contrast, others have not shown an impact of ASPs on antimicrobial resistance.26,30 Cook et al26 reported the successful implementation of an antimicrobial management program, but it did not have an effect on antibiotic resistance. Nevertheless, there is strong evidence to support the causal association of antimicrobial use and antimicrobial resistance.6 Changing rates of antibiotic resistance once they are established might be difficult to accomplish by implementing antimicrobial stewardship practices if they are not sustained over time. Integrating infection-control and antimicrobial stewardship efforts is crucial for the success of controlling the emergence and spread of multidrug-resistant organisms in the hospital setting.7
A limitation of our study is that interventions were performed at a single children's hospital. It is likely that the successful performance of an ASP at a single site depended on customized benchmark reports, order and care sets, and the flexibility of the CPOE to incorporate indications as mandatory fields for each antimicrobial prescribed for all hospitalized children. These implementations and outcomes might not be extrapolated to other academic and nonacademic children's hospitals or to adult hospitals with pediatric wards that use different clinical information systems or electronic medical records. As we previously reported, performance of CPOE systems varies widely, especially among children's hospitals, because these programs have been designed for adult populations and require adjustments when implemented for hospitalized children.11,12
Another important limitation is the lack of a standardized metric to measure antimicrobial use in children. In the early 1980s, the World Health Organization adopted and recommended the use of defined daily doses (DDDs) to standardize and measure antimicrobial use.31 The DDDs are targeted to the average maintenance dose per day for a specific antimicrobial drug per adult patient and does not take into consideration differences in dose-range and weight-based dosing. Even when considered the benchmark standard for adult patients, DDDs cannot be easily applied to measuring antimicrobial use in children.32,33 Alternative metrics, including days of therapy and doses administered, have not been fully evaluated, but for pediatric patients, the number of doses administered reflects antimicrobial use more accurately by taking into account dosage recommendations that vary on the basis of age and clinical diagnosis.31,–,33 Because Gram-negative bacilli susceptibilities remained stable overtime, we could not determine a correlation between the number of doses administered and antibiotic resistance. The 2 main limitations of using the number of doses administered are the difficulties of measuring doses administered without computerized pharmacy records and an overestimation of antimicrobial use when compared with adult patients, because younger children are dosed more frequently than adults.
In our program, interventions were evaluated by an infectious-disease clinical pharmacist and the medical director of the ASP, whereas prior authorization of restricted antimicrobials was granted or denied by the attending infectious-disease physician. In other studies, the appropriateness of antimicrobial recommendations was found to be superior when decisions were made by a team of an infectious-disease pharmacist and attending physician rather than by fellows.34 Differences were attributed to the expertise and trust provided by the medical director of the ASP and accountability of the pharmacist.34 Furthermore, fellows may have perceived that denying these requests might interfere with daily interactions with their colleagues.34 Barlam and DiVall25 reported that in only 1 of 15 top-rated US hospitals were staff infectious-disease physicians responsible for prior authorization of restricted antimicrobials. Results of ASP interventions and outcomes might differ on the basis of the strategies selected. Our strategies were selected on the basis of the characteristics of our patient population, our information technology and electronic medical record capabilities, and human resources available at our institution.
As global commitments and efforts to develop new antibiotics call for multidisciplinary participation, implementation of antimicrobial strategies that target unnecessary antimicrobial use can provide local solutions to the international race against antimicrobial resistance. In our experience, implementation of antimicrobial stewardship strategies had a significant impact on reducing antimicrobial use and improving quality of care of the hospitalized children in our hospital. This successful experience at our children's hospital should encourage other pediatric centers to pursue similar programs.
We acknowledge the helpful comments of Kathryn Edwards, MD, and Terence Dermody, MD, in reviewing the manuscript. We also thank Richard Clapp, BSEE, MA Math (strategic analytics and senior data research and report specialist, Nemours and Alfred I. duPont Hospital for Children) for his ongoing analytical and report-writing assistance.
Implementation of the ASP was supported in part by a grant from the Nemours Clinical Management Program (Orlando, FL).
- Accepted August 31, 2011.
- Address correspondence to M. Cecilia Di Pentima, MD, MPH, Infectious Diseases Division, Department of Pediatrics, Vanderbilt University, D-7235 Medical Center North, 1161 21st Ave South, Nashville, TN 37232-2581. E-mail:
This work was presented in part at the 18th annual meeting of the Society of Health Care Epidemiology; April 5–6, 2008; Orlando, FL (abstract No. 122; poster No. 41).
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
- ASP —
- antimicrobial stewardship program
- CPOE —
- computerized prescriber order entry
- Spellberg B,
- Guidos R,
- Gilbert D,
- et al.
- Boucher HW,
- Talbot GH,
- Bradley JS,
- et al
Infectious Diseases Society of America. The 10 × '20 initiative: pursuing a global commitment to develop 10 new antibacterial drugs by 2020. Clin Infect Dis. 2010;50(8):1081–1083
- Talbot GH,
- Bradley J,
- Edwards JE Jr.,
- Gilbert D,
- Scheld M,
- Bartlett JG
- Owens RC Jr.,
- Shorr AF,
- Deschambeault AL
- Dellit TH,
- Owens RC,
- McGowan JE Jr.,
- et al.
- Paterson DL
- MacDougall C,
- Polk RE
- Di Pentima MC,
- Chan S,
- Eppes SC,
- Klein JD
- Agwu AL,
- Lee CK,
- Jain SK,
- et al
- Berrouane YF,
- Herwaldt LA,
- Pfaller MA
- Coulter M,
- Hossain J,
- Chan S,
- Di Pentima MC
- Di Pentima MC,
- Chan S
- Cook PP,
- Catrou PG,
- Christie JD,
- Young PD,
- Polk RE
- Di Pentima MC,
- Chan S
- Di Pentima MC,
- Chan S
- Pakyz AL,
- Oinonen M,
- Polk RE
- Monnet DL
- Berrington A
- Polk RE,
- Fox C,
- Mahoney A,
- Letcavage J,
- MacDougall C
- Gross R,
- Morgan AS,
- Kinky DE,
- Weiner M,
- Gibson GA,
- Fishman NO
- Copyright © 2011 by the American Academy of Pediatrics