OBJECTIVES:

Identify diagnoses with the highest likelihood of prompting antimicrobial stewardship program (ASP) recommendations and lowest probability of recommendation acceptance, investigate the impact of provider years in practice on recommendation receipt and acceptance, and simultaneously assess the influence of patient and provider-level variables associated with recommendations within a nonfreestanding children’s hospital.

METHODS:

Retrospective cohort study of antibiotic courses reviewed by the ASP staff from December 1, 2014 to November 30, 2016. Poisson regression was used to detect associations between diagnoses, provider years in practice, and the probability of recommendation receipt and acceptance. Multivariable logistic regression was used to simultaneously examine the influence of patient and provider-level characteristics on recommendation probability.

RESULTS:

A total of 938 inpatient encounters and 1170 antibiotic courses were included. Diagnoses were associated with provider receipt (P < .001) and acceptance (P < .001) of recommendations, with ear, nose, and throat and/or sinopulmonary diagnoses most likely to prompt recommendations (56%; 95% confidence interval [CI], 48–64) and recommendations for neonatal and/or infant diagnoses accepted least often (67%; 95% CI, 58–76). No associations were initially found between provider experience and recommendation receipt or acceptance, although multivariable analysis revealed a trend between increasing years in practice and recommendation likelihood (P = .001). Vancomycin usage (64%; 95% CI, 56–72) and ear, nose, and throat and/or sinopulmonary diagnoses (56%; 95% CI, 47–65) had the highest probability of a recommendation. Sensitivity analyses revealed that use of diagnosis-related clinical practice guidelines decreased recommendations and increased acceptance rates, especially for the surgery diagnosis category.

CONCLUSIONS:

High-yield targets for ASP activities at our nonfreestanding children’s hospital were identified. Clinical practice guidelines have the potential to decrease ASP workload, and their development should be particularly encouraged for ASPs with limited resources.

The number of inpatient pediatric antimicrobial stewardship programs (ASPs) throughout the United States continues to increase, secondary to factors including the Centers for Disease Control and Prevention 2014 recommendation for all hospitals to implement ASPs and the Joint Commission’s 2017 stewardship standard requirement for all critical access hospitals.14  Despite this growth, there remains a paucity of published literature from nonfreestanding children’s hospitals used to identify ideal targets for concerted ASP efforts at such institutions.57  Both case mix and patient illness severity at nonfreestanding hospitals differ relative to their freestanding counterparts.8  A greater incidence of patients at freestanding centers undergoes cardiac surgery, chest tube placement, and arterial catheter placement and receives mechanical ventilation.8  Greater relative percentages of patients at freestanding hospitals carry infectious, oncologic, and cardiovascular diagnoses, whereas endocrinologic, gastrointestinal (GI), and injury-related diagnoses account for greater relative percentages at nonfreestanding institutions.8  Clinical and professional responsibilities of physicians in specialties such as hospital medicine can vary between those employed at freestanding versus nonfreestanding facilities.9 

Our previous analysis of a nonfreestanding children’s hospital ASP revealed that reviews of intravenous (IV) vancomycin were most likely to trigger recommendations.7  Patient diagnoses were not assessed during that study, however. Given the known differences between freestanding and nonfreestanding hospitals with regard to patient-level characteristics, variations in patient volume, and provider responsibilities,5,810  our primary objectives of this study were to identify patient diagnoses associated with the highest likelihood of prompting ASP recommendations and lowest likelihood of providers accepting recommendations at our nonfreestanding children’s hospital. We hypothesized that certain diagnoses would be associated with greater probabilities of provider recommendation receipt and lesser likelihoods of acceptance relative to others. Secondary aims included evaluating the impact of provider years in clinical practice on recommendation receipt and acceptance. Similar to Bio et al,11  we anticipated that clinically experienced physicians were most likely to have received recommendations and least likely to have accepted recommendations. In addition, we sought to simultaneously assess patient-level (diagnosis) and provider-level (antibiotic choice, provider team specialty, and years in practice) variables to determine which factors were most predictive of ASP recommendations.

Baystate Children’s Hospital (BCH) is a 110-bed and 57-bassinet nonfreestanding children’s hospital located within Baystate Medical Center (BMC), a 716-bed independent academic medical center in Springfield, Massachusetts. BCH is administratively nonfreestanding because the executive leadership for BMC and BCH is one and the same. BCH contains a 55-bed level III NICU and 7-bed level II PICU, is affiliated with the University of Massachusetts Medical School–Baystate, and serves as the primary teaching facility for 59 resident physicians in categorical pediatrics and internal medicine and/or pediatrics training programs. The University of Massachusetts Medical School–Baystate Department of Pediatrics at BCH is composed of >100 clinical faculty members representing general pediatrics and multiple pediatric subspecialties. In addition to inpatients on the general pediatrics ward managed by an active pediatric hospitalist service, pediatricians from 5 private practices throughout the metropolitan area serve as primary inpatient providers for hospitalized children from their practices. Residents care for patients with hospitalist and private practice providers and place most orders for antibiotics throughout BCH. In general, however, resident antimicrobial prescribing autonomy is tightly restricted for patients with hospitalist attending physicians and for those in the ICUs (although less so for children with private practice providers). A single advanced practitioner is employed part-time for the surgery service, whereas multiple advanced practitioners care for neonates under the close supervision of an attending neonatologist. No bone marrow transplants, solid organ transplants, or pediatric cardiovascular surgical procedures are performed at BCH. Institutional clinical practice guidelines (CPGs) exist for management of (1) febrile infants ≤3 months (which recommend empirical use of ampicillin and gentamicin), (2) children with acute abdominal pain and suspected appendicitis (within which ceftriaxone and metronidazole are advised for complicated appendicitis), and (3) initial dosing of IV vancomycin. No CPGs exist for empirical antibiotic management of inpatients with any other infectious diagnoses such as community-acquired pneumonia (CAP), osteoarticular infections, or skin and/or soft tissue infections.

Established in 2014, the BCH ASP is staffed jointly by infectious diseases physicians and clinical pharmacy specialists. The physician director has 0.1 full-time equivalent salary support dedicated to stewardship, whereas the unit-based pediatric clinical pharmacy specialists have no dedicated full-time equivalent salary support for stewardship. The ASP staff reviews usage of certain predefined antibiotics 48 to 72 hours after initiation (Table 1) for all inpatients with primary inpatient providers (regardless of specialty) who predominantly care for neonates, children, and/or adolescents. The program uses a thrice-weekly prospective audit-with-feedback methodology, a strategy successfully employed at freestanding institutions and nonfreestanding children’s hospitals within larger medical centers.57,1215  ASP recommendations include 8 discrete types (Table 1). All data regarding antibiotics reviewed, recommendations suggested, primary team specialty, provider acceptance of recommendations, and provider implementation of accepted recommendations are collected and recorded in an internal ASP quality improvement database as previously described elsewhere.7  For each date on which stewardship activities occur, all reviews performed are recorded on detailed rounding sheets shared among ASP team members.

TABLE 1

Characteristics of ASP Reviews by Antibiotic, N = 1170

Antibiotic Reviews, n (%)
Antibiotic revieweda  
 Gentamicin 405 (34.6) 
 Ceftriaxone 237 (20.3) 
 Metronidazole (IV and PO) 120 (10.3) 
 Vancomycin (IV) 96 (8.2) 
 Clindamycin (IV and PO) 86 (7.4) 
 Piperacillin-tazobactam 63 (5.4) 
 Cefepime 62 (5.3) 
 Ampicillin-sulbactam 34 (2.9) 
 Levofloxacin (IV and PO) 32 (2.7) 
 Amoxicillin-clavulanate 22 (1.9) 
 Tobramycin (IV) 7 (0.6) 
 Cefotaxime 6 (0.5) 
Recommendation type  
 No recommendation suggestedb 756 (64.6) 
 Discontinue antibiotic 183 (15.6) 
 Recommend formal infectious diseases service consultation 51 (4.4) 
 Dosing adjustment 49 (4.2) 
 Narrow coverage 39 (3.3) 
 Transition from IV to PO 34 (2.9) 
 Broaden coverage 27 (2.3) 
 Toxicity monitoring 22 (1.9) 
 Eliminate redundant antibiotic coverage 9 (0.8) 
Primary inpatient care team  
 NICU 384 (32.8) 
 Hospitalist 264 (22.6) 
 Surgery 180 (15.4) 
 PICU 112 (9.6) 
 Private patient 97 (8.3) 
 Hematology and/or oncology 78 (6.7) 
 Newborn nursery 37 (3.2) 
 Gastroenterology 18 (1.5) 
Diagnosis category  
 Neonatal and/or infant 427 (36.5) 
 Surgery 210 (17.9) 
 ENT and/or sinopulmonary 199 (17.0) 
 Hematologic and/or CNS 139 (11.9) 
 GI and/or genitourinary 111 (9.5) 
 Skin and/or soft tissue 44 (3.8) 
 2 diagnoses 40 (3.4) 
Provider years in clinical practice, y  
 <5 199 (17.0) 
 5–15 534 (45.6) 
 >15 437 (37.4) 
CPGc  
 No 984 (84.1) 
 Yes 186 (15.9) 
Antibiotic Reviews, n (%)
Antibiotic revieweda  
 Gentamicin 405 (34.6) 
 Ceftriaxone 237 (20.3) 
 Metronidazole (IV and PO) 120 (10.3) 
 Vancomycin (IV) 96 (8.2) 
 Clindamycin (IV and PO) 86 (7.4) 
 Piperacillin-tazobactam 63 (5.4) 
 Cefepime 62 (5.3) 
 Ampicillin-sulbactam 34 (2.9) 
 Levofloxacin (IV and PO) 32 (2.7) 
 Amoxicillin-clavulanate 22 (1.9) 
 Tobramycin (IV) 7 (0.6) 
 Cefotaxime 6 (0.5) 
Recommendation type  
 No recommendation suggestedb 756 (64.6) 
 Discontinue antibiotic 183 (15.6) 
 Recommend formal infectious diseases service consultation 51 (4.4) 
 Dosing adjustment 49 (4.2) 
 Narrow coverage 39 (3.3) 
 Transition from IV to PO 34 (2.9) 
 Broaden coverage 27 (2.3) 
 Toxicity monitoring 22 (1.9) 
 Eliminate redundant antibiotic coverage 9 (0.8) 
Primary inpatient care team  
 NICU 384 (32.8) 
 Hospitalist 264 (22.6) 
 Surgery 180 (15.4) 
 PICU 112 (9.6) 
 Private patient 97 (8.3) 
 Hematology and/or oncology 78 (6.7) 
 Newborn nursery 37 (3.2) 
 Gastroenterology 18 (1.5) 
Diagnosis category  
 Neonatal and/or infant 427 (36.5) 
 Surgery 210 (17.9) 
 ENT and/or sinopulmonary 199 (17.0) 
 Hematologic and/or CNS 139 (11.9) 
 GI and/or genitourinary 111 (9.5) 
 Skin and/or soft tissue 44 (3.8) 
 2 diagnoses 40 (3.4) 
Provider years in clinical practice, y  
 <5 199 (17.0) 
 5–15 534 (45.6) 
 >15 437 (37.4) 
CPGc  
 No 984 (84.1) 
 Yes 186 (15.9) 

CNS, central nervous system; PO, oral.

a

IV amikacin was also eligible for review; however, no amikacin reviews were performed during the study time period.

b

Because of ASP agreement with primary inpatient care team management plan.

c

Febrile infants ≤3 mo and gangrenous or perforated appendicitis.

A retrospective cohort study was performed for all BCH inpatients from December 1, 2014 to November 30, 2016, whose antibiotic courses of therapy underwent review as per the internal ASP database. The BMC Institutional Review Board reviewed and approved all facets of this project.

Working Diagnosis Categories

Copies of the detailed rounding sheets were reviewed to ascertain patient working diagnoses at the time of ASP review. A patient’s working diagnosis was defined as the presumptive clinical condition (per the patient’s primary provider) that necessitated antibiotic use at the time ASP review occurred. For reviews performed for which no working diagnoses were recorded on the rounding sheets, chart reviews of provider documentation on the dates of ASP reviews were used to establish diagnoses. After identification of all working diagnoses, the diagnoses were condensed into 7 categories by using an approach similar to that used by Goldman et al16  (Table 1). Included among these was a category for patients for whom >1 working diagnosis category was applicable (ie, 2 diagnoses).

Provider Years in Clinical Practice

A list was generated of all potential BCH inpatient primary providers from December 1, 2014 to November 30, 2016. Board examination certification websites (www.abp.org and www.absurgery.org) were reviewed to obtain initial certification dates for potential providers, which were then used to calculate years in clinical practice. For subspecialists actively practicing in their certified subspecialty during the study period, initial dates of subspecialty certification were used to determine years in clinical practice. Providers were categorized into 1 of 3 groups based on corresponding length of time between initial certification date and December 1, 2014 (i.e., <5, 5–15, and >15 years). Providers on the dates on which ASP reviews occurred were determined via chart review.

Descriptive statistics were used to summarize characteristics at the patient review level (ie, characteristics of the patient at the time of the ASP review) and the antibiotic level (ie, characteristics of the antibiotic prompting the review). Study outcomes included the probability of ASP recommendation and, among those with a recommendation, the probability of acceptance. By using separate models, we estimated these probabilities for 2 primary predictor variables of interest: diagnosis category and provider years in clinical practice. An overall Wald test was used to test the significance of these predictors in the model, and linear contrasts were used to test the trend of increasing years in practice. All tests used a 2-sided α of .05. One of the features of this data set is that providers were repeatedly observed over the study duration. If providers either prescribed a certain way (eg, in a manner prompting more ASP reviews relative to their peers) and/or consistently accepted or rejected ASP recommendations over time, they could unduly weight our estimates. To account for those possibilities, separate generalized estimating equation Poisson regression models clustering on provider were used to generate population-averaged estimates and 95% confidence intervals (CIs). To estimate probabilities (ie, average marginal effects), we used Stata’s (Stata Corp, College Station, TX) margins command. In terms of power, for an overall χ2-type analysis on a 7 × 2 table (eg, 7 diagnostic categories by presence or absence of a recommendation) with 6 degrees of freedom, we had sufficient power to detect a small effect size (Cohen’s w)17  of 0.125 for an anticipated sample size of 882 reviews.

To address which patient-level and/or provider-level variables were most predictive of ASP recommendations, we created a multivariable generalized estimating equation logistic regression model (also clustering on provider) to simultaneously estimate adjusted recommendation probabilities and 95% CIs for each diagnosis category, provider years in clinical practice, primary team specialty, and antibiotic reviewed.

We also performed multiple sensitivity analyses. Because the presence of CPGs could impact the probability of a recommendation and/or acceptance, we restricted all of the models (unadjusted and multivariable) to ASP reviews without a CPG and assessed the results for clinically important differences. For the sensitivity analyses, we chose to exclude only the 2 diagnosis-related CPGs (ie, gangrenous and/or perforated appendicitis and febrile infants ≤3 months) given the incomplete overlap of the CPG with the diagnosis categories, whereas vancomycin estimates represent the initial dosing CPG in all cases. All analysis was conducted by using Stata MP, version 15.1.

From December 1, 2014 to November 30, 2016, the ASP staff reviewed 1170 discrete antibiotic courses prescribed during 938 individual patient encounters (Table 1). Overall, 414 of 1170 (35%) reviews prompted ASP recommendations. Providers accepted 307 of 414 (74%) recommendations, with 287 of 307 (93%) ultimately implemented.

Specific diagnoses within each category are listed in Table 2. There was a significant association found between working diagnosis category and probability of recommendation (P < .001) (Fig 1). The ear, nose, and throat (ENT) and/or sinopulmonary category was associated with the greatest likelihood of triggering recommendations (56%; 95% CI, 48–64), with antibiotic discontinuation (n = 41; 35% of all ENT and/or sinopulmonary recommendations) and narrowing coverage (n = 21; 18%) advised most often. The surgery category was least likely to receive recommendations (23%; 95% CI, 14–32), with all remaining probabilities ranging between 28% and 37%. Of note, a total of 31 of 427 antibiotic courses reviewed for neonatal and/or infant diagnoses involved a CPG for febrile infants ≤3 months. Of those, 12 (39%) resulted in a recommendation. Additionally, 155 of 210 antibiotic courses reviewed for surgery diagnoses involved a CPG for gangrenous and/or perforated appendicitis, with 16 (10%) resulting in a recommendation. After restricting the analysis to ASP reviews without a CPG, ASP recommendations for surgery diagnoses increased from 23% to 38%, whereas the rest remained relatively stable. ENT and/or sinopulmonary diagnoses continued to have the highest probability (58%), whereas neonatal and/or infant diagnoses had the lowest probability (30%).

TABLE 2

Working Diagnoses and Their Categories by Patient, N = 938

Diagnosis Category and Diagnosisn%
GI and/or genitourinary   
 Overall 93 — 
 Urinary tract infection 41 44.1 
 Pyelonephritis 20 21.5 
 Inflammatory bowel disease exacerbation 10 10.8 
 Abscess(es) due to inflammatory bowel disease 7.5 
 Clostridium difficile infection 6.5 
 Other 9.7 
ENT and/or sinopulmonary   
 Overall 159 — 
 CAP 46 28.9 
 Aspiration pneumonia 28 17.6 
 2 ENT and/or sinopulmonary diagnoses 17 10.7 
 Tracheitis 14 8.8 
 CF exacerbation 11 6.9 
 Peritonsillar and/or retropharyngeal abscess 5.0 
 Hospital-acquired pneumonia 4.4 
 Dental abscess 3.8 
 Ventilator-associated pneumonia 3.8 
 Adenitis ± abscess 3.1 
 Other 11 6.9 
Neonatal and/or infant   
 Overall 388 — 
 Suspected or confirmed early-onset neonatal sepsis 244 62.9 
 Meconium aspiration 33 8.5 
 Febrile infant ≤3 mo 30 7.7 
 Early-onset neonatal sepsis plus neonatal pneumonia 27 7.0 
 Necrotizing enterocolitis 25 6.4 
 Suspected or confirmed late-onset neonatal sepsis 15 3.9 
 Neonatal pneumonia 12 3.1 
 Other 0.5 
Surgery   
 Overall 123 — 
 Gangrenous and/or perforated appendicitis 82 66.7 
 Intraabdominal process unrelated to perforated appendicitis 17 13.8 
 Postoperative management 11 8.9 
 Surgical site infection 7.3 
 Other 3.3 
Skin and/or soft tissue     
 Overall 38 — 
 Skin and/or soft tissue infection 37 97.4 
 Omphalitis 2.6 
Hematologic and/or CNS   
 Overall 113 — 
 Fever and/or neutropenia in oncology patient 48 42.5 
 Sickle cell anemia exacerbation or suspected acute chest syndrome 27 23.9 
 Bacteremia 26 23.0 
 Hypothermia 4.4 
 Other 6.2 
2 distinct diagnoses 24 — 
Diagnosis Category and Diagnosisn%
GI and/or genitourinary   
 Overall 93 — 
 Urinary tract infection 41 44.1 
 Pyelonephritis 20 21.5 
 Inflammatory bowel disease exacerbation 10 10.8 
 Abscess(es) due to inflammatory bowel disease 7.5 
 Clostridium difficile infection 6.5 
 Other 9.7 
ENT and/or sinopulmonary   
 Overall 159 — 
 CAP 46 28.9 
 Aspiration pneumonia 28 17.6 
 2 ENT and/or sinopulmonary diagnoses 17 10.7 
 Tracheitis 14 8.8 
 CF exacerbation 11 6.9 
 Peritonsillar and/or retropharyngeal abscess 5.0 
 Hospital-acquired pneumonia 4.4 
 Dental abscess 3.8 
 Ventilator-associated pneumonia 3.8 
 Adenitis ± abscess 3.1 
 Other 11 6.9 
Neonatal and/or infant   
 Overall 388 — 
 Suspected or confirmed early-onset neonatal sepsis 244 62.9 
 Meconium aspiration 33 8.5 
 Febrile infant ≤3 mo 30 7.7 
 Early-onset neonatal sepsis plus neonatal pneumonia 27 7.0 
 Necrotizing enterocolitis 25 6.4 
 Suspected or confirmed late-onset neonatal sepsis 15 3.9 
 Neonatal pneumonia 12 3.1 
 Other 0.5 
Surgery   
 Overall 123 — 
 Gangrenous and/or perforated appendicitis 82 66.7 
 Intraabdominal process unrelated to perforated appendicitis 17 13.8 
 Postoperative management 11 8.9 
 Surgical site infection 7.3 
 Other 3.3 
Skin and/or soft tissue     
 Overall 38 — 
 Skin and/or soft tissue infection 37 97.4 
 Omphalitis 2.6 
Hematologic and/or CNS   
 Overall 113 — 
 Fever and/or neutropenia in oncology patient 48 42.5 
 Sickle cell anemia exacerbation or suspected acute chest syndrome 27 23.9 
 Bacteremia 26 23.0 
 Hypothermia 4.4 
 Other 6.2 
2 distinct diagnoses 24 — 

CF, cystic fibrosis; CNS, central nervous system; —, not applicable.

FIGURE 1

Probabilities of provider receipt and acceptance of ASP recommendations by working diagnosis category. CNS, central nervous system.

FIGURE 1

Probabilities of provider receipt and acceptance of ASP recommendations by working diagnosis category. CNS, central nervous system.

Close modal

Among 414 recommendations proposed, a significant association was found between the working diagnosis category and probability of provider recommendation acceptance (P < .001) (Fig 1). Acceptance was 100% for patients with skin and/or soft tissue infectious diagnoses; however, only 15 total interventions were suggested for that category. Acceptance was also high for the GI and/or genitourinary category (89%; 95% CI, 80–98), with 38 of 43 recommendations accepted. The neonatal and/or infant category had the lowest acceptance rate (67%; 95% CI, 58–76), with 94 of 140 suggestions accepted. Of the 46 rejected neonatal and/or infant category recommendations, 34 (74%) advised antibiotic discontinuation. Diagnoses for 32 of 34 (94%) included ≥1 of the following: suspected and/or confirmed early-onset neonatal sepsis, neonatal pneumonia, and/or necrotizing enterocolitis. For all remaining diagnosis categories, acceptance ranged from 71% to 76%. After restricting the analysis to ASP reviews without a CPG, acceptance remained highest for GI and/or genitourinary (89%) and skin and/or soft tissue diagnoses (100%). Acceptance probabilities for neonatal and/or infant diagnoses decreased slightly (to 64%), whereas acceptance for surgery diagnoses decreased from 71% to 60% (making surgery the category with the lowest acceptance rate).

Of the 92 providers whose patients underwent ASP review, 15 (16%) were in clinical practice <5 years, with 31 (34%) between 5 and 15 years and 46 (50%) for >15 years. Although likelihood of a recommendation was lowest among those with <5 years’ experience (24% vs 39% [5–15] and 35% [>15]), this was not statistically significant (P = .24) and no trend was observed (Ptrend = .18). Acceptance was slightly higher among those with 5 to 15 years of experience (78% vs ∼70% in the other 2 groups); however, this also was not statistically significant (P = .36) and no trend was observed (Ptrend = .94) (Fig 2). Inferences for both likelihood of a recommendation and acceptance remained the same after restricting to ASP reviews without a CPG.

FIGURE 2

Probabilities of provider receipt and acceptance of ASP recommendations by provider years in clinical practice.

FIGURE 2

Probabilities of provider receipt and acceptance of ASP recommendations by provider years in clinical practice.

Close modal

When simultaneously taking into consideration patient-level (diagnosis) and provider-level (antibiotic choice, provider team specialty, and years in practice) factors, the variable with the highest adjusted probability of ASP recommendation was IV vancomycin use (64%; 95% CI, 56–72) (Fig 3). Of the 63 vancomycin recommendations, 9 (14%) suggested dosing adjustment or toxicity monitoring, whereas 42 (67%) advised narrowing coverage or discontinuation. The ENT and/or sinopulmonary diagnosis category was also strongly predictive of ASP intervention (56%; 95% CI, 47–65). Although having a gastroenterologist primary team provider was associated with a 53% probability of ASP recommendation, gastroenterologists prescribed only 18 courses of reviewed antibiotics (during 11 individual patient encounters), resulting in a wide CI and therefore unstable estimate. A significant trend across all provider years in practice groups was identified, with recommendation probabilities increasing according to the duration of clinical experience (26% [<5 years] to 35% [5–15 years] and finally 39% [>15 years]) (Ptrend = .001). After restricting the multivariable model to ASP reviews without a diagnosis-related CPG, we continued to observe that IV vancomycin had the highest probability of a recommendation (increased to 66%) followed by ENT and/or sinopulmonary diagnoses (increased to 61%).

FIGURE 3

Results of a multivariable assessment of provider, patient, and antimicrobial characteristics and their relative probabilities of prompting ASP recommendations. CNS, central nervous system; PO, oral.

FIGURE 3

Results of a multivariable assessment of provider, patient, and antimicrobial characteristics and their relative probabilities of prompting ASP recommendations. CNS, central nervous system; PO, oral.

Close modal

Patient working diagnosis categories at the time of ASP review significantly influenced probabilities of provider receipt and acceptance of ASP recommendations at our nonfreestanding children’s hospital. No significant associations were initially detected between provider years in clinical practice and recommendation receipt or acceptance, although the multivariable assessment revealed a trend that providers with >15 years in practice were most likely to receive a recommendation. Despite this observed trend, we also found through multivariable analysis that provider years in practice was not among those variables most likely to prompt a recommendation. Our multivariable assessment results strongly emphasize the need for measures to assist providers with the selection of ideal empirical antibiotic therapy for ENT and/or sinopulmonary infections and reinforce the conclusion from previous studies that IV vancomycin usage optimization should serve as a high-yield target for stewardship programs regardless of institutional size or breadth of clinical practice.6,7,13,1821  The restricted analyses revealed that the presence of institutional CPGs may have an important effect in curbing the need for certain ASP recommendations.

Of all the working diagnosis categories, the ENT and/or sinopulmonary category had the strongest association with ASP recommendations (with 53% advising antibiotic discontinuation or narrowing coverage). This finding is consistent with published data from freestanding children’s hospitals used to assert that diagnoses like CAP are likely to trigger recommendations to discontinue and/or modify antibiotic therapy.16,22,23  We support the previously stated claim that ASP optimization of antibiotic therapy for diagnoses including sinopulmonary infections like CAP at nonfreestanding children’s hospitals is necessary at a relative frequency equal to or perhaps even greater than that encountered at freestanding institutions.10  Our findings are particularly relevant because pneumonia is known to be the most costly diagnosis for pediatric hospitals that are not freestanding (in contrast to freestanding centers, where costs are greatest for respiratory distress syndrome in newborns and chemotherapy admissions).24  Recommendation acceptance was high for patients in the GI and/or genitourinary category (nearly 66% of which had a urinary tract infection or pyelonephritis), likely because those recommendations were based primarily on urine culture isolate susceptibility testing results that typically returned 48 to 72 hours after being obtained (ie, a duration of time coinciding closely with ASP antimicrobial review).

Providers of patients with neonatal and/or infant category diagnoses had the lowest likelihood of recommendation acceptance in the unrestricted analysis. Most were neonatologists, and obstacles to successful antimicrobial stewardship in the NICU have been well documented in the medical literature.2527  The majority of providers who rejected neonatal and/or infant category recommendations advised discontinuation of antibiotics initiated empirically for conditions including early-onset neonatal sepsis and necrotizing enterocolitis. Standardization of the management approaches to such conditions is likely to provide significant benefit.

Rejection of ASP recommendations has previously been linked to behaviors of experienced clinicians relying on past practices that may have resulted in positive outcomes, and such physicians may be more skeptical of new stewardship interventions relative to their less experienced colleagues.28,29  Bio et al11  did not initially observe any differences between the duration of provider experience and ASP recommendation acceptance after univariate analysis, although they found a small but statistically significant difference between those variables in their adjusted multivariate model. Similarly, our study did not initially reveal any associations between provider years in clinical practice and likelihood of ASP recommendation receipt. However, when exploring the relationship between those variables within the context of the multivariable analysis, a significant trend between provider years in practice and recommendations emerged, which reveals that ASP recommendations were most common among experienced providers. Although this finding could be an anomaly unique to BCH, the observed similarities between our results and those of Bio et al11  suggest that the impact of provider clinical experience on ASP recommendations is closely tied to additional factors, including patient diagnosis and primary team specialty. Although provider years in practice was not among the factors most predictive of recommendations in our multivariable analysis, the observed trend between increasing provider years in practice and recommendations reveals that the most effective stewardship educational strategies will account for multiple patient-level and provider-level factors.

The multivariable assessment also revealed that IV vancomycin use had the greatest association with ASP recommendation receipt. A substantial majority of vancomycin recommendations made during the study period were for de-escalation or discontinuation, whereas only a small number were for vancomycin dosing adjustment and/or toxicity monitoring. Such results reflect the need for provider education regarding appropriate indications for vancomycin initiation and discontinuation at BCH. The multivariable results likewise reveal the conclusion that BCH patients with ENT and/or sinopulmonary diagnoses frequently trigger ASP recommendations.

The impact of diagnosis-related CPGs on BCH stewardship activities was varied. Among neonatal and/or infant diagnoses, the probability of a recommendation for febrile infants ≤3 months (39%) was actually higher than the probability for the overall diagnosis category (28%). However, that specific diagnosis accounted for only a small percentage of all ASP reviews performed for patients with neonatal or infant diagnoses. When restricting our analyses to no CPGs, we observed no substantive differences in the probability of a recommendation or acceptance for that diagnosis category. In contrast, reviews of gangrenous and/or perforated appendicitis comprised a large percentage of all surgery category reviews performed, and the probability of a recommendation for that diagnosis (10%) was much lower than for the overall diagnosis category (23%). Furthermore, when restricting our analyses to no CPGs, we observed not only that surgery category recommendations increased by 15% but also that recommendation acceptance decreased by 11%. We believe that these findings reveal the importance of CPGs for the management of pediatric appendicitis, as previously stated by Willis et al.30,31  Because the BCH CPG for management of patients with appendicitis predated ASP establishment, we, on the basis of our finding, could also imply that CPGs for select diagnoses at nonfreestanding children’s hospitals are equally as important as (or perhaps even more important than) the performance of regular prospective audit with feedback for patients with those same diagnoses.

This study has multiple limitations. The study was performed at a single nonfreestanding children’s hospital. The case mix of children and frequencies with which certain diagnoses are seen at BCH could differ from other nonfreestanding facilities that offer varying ranges of clinical services. The precise impact of CPGs on ASP recommendations may be under- or overestimated given the provider and case mix at BCH. Despite participation during the study period in a multicenter collaborative aimed at optimizing antibiotic use for CAP,32  BCH does not have a CPG for management of that condition. Such guidelines have previously been shown to decrease broad-spectrum antimicrobial usage at freestanding children’s hospitals, and their development is encouraged by the Infectious Diseases Society of America.12,3234  The influence of initial dosing CPGs for vancomycin may be underestimated by us in this study because most of the recommendations for vancomycin were for de-escalation or discontinuation (whereas the BCH CPG is focused on optimal initial dosing of vancomycin). Regular ASP monitoring of a predefined list of antibiotics could lead to repeated review of specific diagnoses to the possible exclusion of conditions for which reviewed antibiotics are not prescribed. However, our list composition closely resembles that of the only other nonfreestanding US children’s hospital besides ours for which such information is currently available in the published medical literature.6  The total number of providers whose patients underwent ASP review was low, and therefore the impact of provider years in practice may be highly variable and dependent on practice patterns of a few physicians. However, one would expect that institutions with a size and scope of practice similar to BCH would have comparable numbers of clinical faculty members. The extent of providers’ exposure to ASPs before interaction with ours (eg, at previous places of employment or during residency and/or fellowship training) was not assessed for this study, and such experiences may have influenced provider decision-making.

The information gleaned from this study reveals an emphasis on patient diagnoses toward which stewardship efforts should be focused at nonfreestanding children’s hospitals similar to BCH. On a broader scale, the procedural analyses of this study can be used by any ASP that performs prospective audit with feedback to inform priority areas regardless of institutional size, setting, or scope of clinical practice. Areas for future emphasis as identified in this study include development of institutional CPGs that dovetail with patient diagnoses and usage of specific antimicrobial agents regularly prompting ASP recommendations. The importance of well-crafted CPGs in decreasing ASP workload cannot be overemphasized. This is particularly relevant for nonfreestanding children’s hospitals similar to ours, within which resources for dedicated ASP support can be lacking relative to programs at freestanding centers (and also when compared with proposed minimum standards necessary to demonstrate ASP effectiveness).4,5,7,35  Finally, ASP team members should be aware that providers’ previous clinical experiences may influence their current clinical practice patterns.

Dr Klatte conceptualized and designed the study, performed data collection, and drafted the initial manuscript; Mr Knee designed the data collection instruments, conducted the initial analyses, and reviewed and revised the manuscript; Drs Szczerba, Horton, and Kopcza performed data collection and reviewed and revised the manuscript; Dr Fisher performed data collection and critically reviewed the manuscript; and all authors approved the final manuscript as submitted.

FUNDING: No external funding.

1
Pollack
LA
,
Srinivasan
A
.
Core elements of hospital antibiotic stewardship programs from the Centers for Disease Control and Prevention
.
Clin Infect Dis
.
2014
;
59
(
suppl 3
):
S97
S100
2
Joint Commission on Hospital Accreditation
.
APPROVED: new antimicrobial stewardship standard
.
Jt Comm Perspect
.
2016
;
36
(
7
):
1, 3
4, 8
3
Godbout
EJ
,
Pakyz
AL
,
Markley
JD
,
Noda
AJ
,
Stevens
MP
.
Pediatric antimicrobial stewardship: state of the art
.
Curr Infect Dis Rep
.
2018
;
20
(
10
):
39
4
McPherson
C
,
Lee
BR
,
Terrill
C
, et al
.
Characteristics of pediatric antimicrobial stewardship programs: current status of the Sharing Antimicrobial Reports for Pediatric Stewardship (SHARPS) collaborative
.
Antibiotics (Basel)
.
2018
;
7
(
1
):
E4
5
Turner
RB
,
Valcarlos
E
,
Loeffler
AM
,
Gilbert
M
,
Chan
D
.
Impact of an antimicrobial stewardship program on antibiotic use at a nonfreestanding children’s hospital
.
J Pediatric Infect Dis Soc
.
2017
;
6
(
3
):
e36
e40
6
Lighter-Fisher
J
,
Desai
S
,
Stachel
A
,
Pham
VP
,
Klejmont
L
,
Dubrovskaya
Y
.
Implementing an inpatient pediatric prospective audit and feedback antimicrobial stewardship program within a larger medical center
.
Hosp Pediatr
.
2017
;
7
(
9
):
516
522
7
Klatte
JM
,
Kopcza
K
,
Knee
A
,
Horton
ER
,
Housman
E
,
Fisher
DJ
.
Implementation and impact of an antimicrobial stewardship program at a non-freestanding children’s hospital
.
J Pediatr Pharmacol Ther
.
2018
;
23
(
2
):
84
91
8
Gupta
P
,
Rettiganti
M
,
Fisher
PL
,
Chang
AC
,
Rice
TB
,
Wetzel
RC
.
Association of freestanding children’s hospitals with outcomes in children with critical illness
.
Crit Care Med
.
2016
;
44
(
12
):
2131
2138
9
Freed
GL
,
Dunham
KM
;
Research Advisory Committee of the American Board of Pediatrics
.
Pediatric hospitalists: training, current practice, and career goals
.
J Hosp Med
.
2009
;
4
(
3
):
179
186
10
Leyenaar
JK
,
Lagu
T
,
Shieh
MS
,
Pekow
PS
,
Lindenauer
PK
.
Variation in resource utilization for the management of uncomplicated community-acquired pneumonia across community and children’s hospitals
.
J Pediatr
.
2014
;
165
(
3
):
585
591
11
Bio
LL
,
Kruger
JF
,
Lee
BP
,
Wood
MS
,
Schwenk
HT
.
Predictors of antimicrobial stewardship program recommendation disagreement
.
Infect Control Hosp Epidemiol
.
2018
;
39
(
7
):
806
813
12
Barlam
TF
,
Cosgrove
SE
,
Abbo
LM
, et al
.
Implementing an antibiotic stewardship program: guidelines by the Infectious Diseases Society of America and the Society for Healthcare Epidemiology of America
.
Clin Infect Dis
.
2016
;
62
(
10
):
e51
e77
13
Hurst
AL
,
Child
J
,
Pearce
K
,
Palmer
C
,
Todd
JK
,
Parker
SK
.
Handshake stewardship: a highly effective rounding-based antimicrobial optimization service
.
Pediatr Infect Dis J
.
2016
;
35
(
10
):
1104
1110
14
Di Pentima
MC
,
Chan
S
,
Hossain
J
.
Benefits of a pediatric antimicrobial stewardship program at a children’s hospital
.
Pediatrics
.
2011
;
128
(
6
):
1062
1070
15
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
(
3
):
179
186
16
Goldman
JL
,
Lee
BR
,
Hersh
AL
, et al
.
Clinical diagnoses and antimicrobials predictive of pediatric antimicrobial stewardship recommendations: a program evaluation
.
Infect Control Hosp Epidemiol
.
2015
;
36
(
6
):
673
680
17
Cohen
J
.
Statistical Power Analysis for the Behavioral Sciences
. 2nd ed.
Hillsdale, NJ
:
Lawrence Erlbaum Associates
;
1988
18
Gillon
J
,
Xu
M
,
Slaughter
J
,
Di Pentima
MC
.
Vancomycin use: room for improvement among hospitalized children
.
J Pharm Pract
.
2017
;
30
(
3
):
296
299
19
Nguyen-Ha
PT
,
Howrie
D
,
Crowley
K
, et al
.
A quality assessment of a collaborative model of a pediatric antimicrobial stewardship program
.
Pediatrics
.
2016
;
137
(
5
):
e20150316
20
Di Pentima
MC
,
Chan
S
.
Impact of antimicrobial stewardship program on vancomycin use in a pediatric teaching hospital
.
Pediatr Infect Dis J
.
2010
;
29
(
8
):
707
711
21
Levy
ER
,
Swami
S
,
Dubois
SG
,
Wendt
R
,
Banerjee
R
.
Rates and appropriateness of antimicrobial prescribing at an academic children’s hospital, 2007-2010
.
Infect Control Hosp Epidemiol
.
2012
;
33
(
4
):
346
353
22
McCulloh
RJ
,
Queen
MA
,
Lee
B
, et al
.
Clinical impact of an antimicrobial stewardship program on pediatric hospitalist practice, a 5-year retrospective analysis
.
Hosp Pediatr
.
2015
;
5
(
10
):
520
527
23
Gerber
JS
,
Kronman
MP
,
Ross
RK
, et al
.
Identifying targets for antimicrobial stewardship in children’s hospitals
.
Infect Control Hosp Epidemiol
.
2013
;
34
(
12
):
1252
1258
24
Leyenaar
JK
,
Ralston
SL
,
Shieh
MS
,
Pekow
PS
,
Mangione-Smith
R
,
Lindenauer
PK
.
Epidemiology of pediatric hospitalizations at general hospitals and freestanding children’s hospitals in the United States
.
J Hosp Med
.
2016
;
11
(
11
):
743
749
25
Cantey
JB
,
Wozniak
PS
,
Pruszynski
JE
,
Sánchez
PJ
.
Reducing unnecessary antibiotic use in the neonatal intensive care unit (SCOUT): a prospective interrupted time-series study
.
Lancet Infect Dis
.
2016
;
16
(
10
):
1178
1184
26
Nzegwu
NI
,
Rychalsky
MR
,
Nallu
LA
, et al
.
Implementation of an antimicrobial stewardship program in a neonatal intensive care unit
.
Infect Control Hosp Epidemiol
.
2017
;
38
(
10
):
1137
1143
27
Cantey
JB
,
Patel
SJ
.
Antimicrobial stewardship in the NICU
.
Infect Dis Clin North Am
.
2014
;
28
(
2
):
247
261
28
Goldstein
EJ
,
Goff
DA
,
Reeve
W
, et al
.
Approaches to modifying the behavior of clinicians who are noncompliant with antimicrobial stewardship program guidelines
.
Clin Infect Dis
.
2016
;
63
(
4
):
532
538
29
Szymczak
JE
,
Feemster
KA
,
Zaoutis
TE
,
Gerber
JS
.
Pediatrician perceptions of an outpatient antimicrobial stewardship intervention
.
Infect Control Hosp Epidemiol
.
2014
;
35
(
suppl 3
):
S69
S78
30
Willis
ZI
,
Duggan
EM
,
Gillon
J
,
Blakely
ML
,
Di Pentima
MC
.
Improvements in antimicrobial prescribing and outcomes in pediatric complicated appendicitis
.
Pediatr Infect Dis J
.
2018
;
37
(
5
):
429
435
31
Willis
ZI
,
Duggan
EM
,
Bucher
BT
, et al
.
Effect of a clinical practice guideline for pediatric complicated appendicitis
.
JAMA Surg
.
2016
;
151
(
5
):
e160194
32
Parikh
K
,
Biondi
E
,
Nazif
J
, et al
;
Value in Inpatient Pediatrics Network Quality Collaborative for Improving Care in Community Acquired Pneumonia
.
A multicenter collaborative to improve care of community acquired pneumonia in hospitalized children
.
Pediatrics
.
2017
;
139
(
3
):
e20161411
33
Smith
MJ
,
Kong
M
,
Cambon
A
,
Woods
CR
.
Effectiveness of antimicrobial guidelines for community-acquired pneumonia in children
.
Pediatrics
.
2012
;
129
(
5
).
34
Newman
RE
,
Hedican
EB
,
Herigon
JC
,
Williams
DD
,
Williams
AR
,
Newland
JG
.
Impact of a guideline on management of children hospitalized with community-acquired pneumonia
.
Pediatrics
.
2012
;
129
(
3
).
35
Doernberg
SB
,
Abbo
LM
,
Burdette
SD
, et al
.
Essential resources and strategies for antibiotic stewardship programs in the acute care setting
.
Clin Infect Dis
.
2018
;
67
(
8
):
1168
1174

Competing Interests

POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.

FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.