Nearly 25% of antibiotics prescribed to children are inappropriate or unnecessary, subjecting patients to avoidable adverse medication effects and cost.
We conducted a quality improvement initiative across 118 hospitals participating in the American Academy of Pediatrics Value in Inpatient Pediatrics Network 2020 to 2022. We aimed to increase the proportion of children receiving appropriate: (1) empirical, (2) definitive, and (3) duration of antibiotic therapy for community-acquired pneumonia, skin and soft tissue infections, and urinary tract infections to ≥85% by Jan 1, 2022. Sites reviewed encounters of children >60 days old evaluated in the emergency department or hospital. Interventions included monthly audit with feedback, educational webinars, peer coaching, order sets, and a mobile app containing site-specific, antibiogram-based treatment recommendations. Sites submitted 18 months of baseline, 2-months washout, and 10 months intervention data. We performed interrupted time series (analyses for each measure.
Sites reviewed 43 916 encounters (30 799 preintervention, 13 117 post). Overall median [interquartile range] adherence to empirical, definitive, and duration of antibiotic therapy was 67% [65% to 70%]; 74% [72% to 75%] and 61% [58% to 65%], respectively at baseline and was 72% [71% to 72%]; 79% [79% to 80%] and 71% [69% to 73%], respectively, during the intervention period. Interrupted time series revealed a 13% (95% confidence interval: 1% to 26%) intercept change at intervention for empirical therapy and a 1.1% (95% confidence interval: 0.4% to 1.9%) monthly increase in adherence per month for antibiotic duration above baseline rates. Balancing measures of care escalation and revisit or readmission did not increase.
This multisite collaborative increased appropriate antibiotic use for community-acquired pneumonia, skin and soft tissue infections, and urinary tract infection among diverse hospitals.
Up to 25% of antibiotics prescribed to US children are either inappropriate or unnecessary,1 –4 subjecting millions of children annually to avoidable and potentially life-long health consequences, adverse medication effects, and healthcare costs.5 Inappropriate antibiotic use promotes the emergence of antimicrobial-resistant organisms6 ; reducing such use may decrease the prevalence of antibiotic-resistant bacteria.7 Community-acquired pneumonia (CAP), skin and soft tissue infections (SSTIs, eg, cellulitis and skin abscesses), and urinary tract infections (UTIs) comprise 3 of the most common infections for which children receive care in the emergency department (ED) or hospital.8 These diagnoses represent nearly 2.5 million ED visits and 250 000 hospitalizations annually in the United States.8
Despite the presence of evidence-based guidelines for managing these infections,9 –11 antibiotic prescription practices vary widely.12 –14 For example, 4 years after guideline publication 39% of children hospitalized for CAP in free-standing children’s hospitals and 73% of children hospitalized in nonchildren’s hospitals receive inappropriate empirical (ie, initial) antibiotic therapy.13 A review of 1319 young infants treated at 36 free-standing children’s hospitals for SSTI found that providers prescribed >130 unique empirical antibiotic combinations.14 A national study of ED prescribing found that 32% of the roughly 7 million antibiotics prescriptions for children were not indicated.15 Prior research also shows that most children receive unnecessarily prolonged (>7 days) antimicrobial therapy. A multisite retrospective study of Medicaid claims data found that >77% of children seen for SSTI received prolonged antibiotic therapy.16
Changing antibiotic use in healthcare organizations requires a multifaceted approach to successfully disseminate recommendations in ways that durably change prescribing patterns. Quality improvement (QI) projects are expensive to conduct and maintain,17 particularly for smaller healthcare organizations that often lack programmatic, personnel, and/or institutional resources available to larger institutions.18 Over two-thirds of hospitalized children receive their care at hospitals that are not free-standing children’s hospitals, including nonfreestanding children’s hospitals and general hospitals.19 General hospitals may use general inpatient beds, pediatric wards, or a children’s hospital nested inside an adult hospital or hospital system. Although free-standing children’s hospitals and many nonfreestanding children’s hospitals maintain pediatric antimicrobial stewardship programs, many general hospitals do not maintain such programs.18 ,20 Quality improvement collaboratives can play an effective role in shared learning and dissemination of best practices. Evidence suggests that such collaboratives can improve antimicrobial use.21 ,22 The American Academy of Pediatrics (AAP) Value in Inpatient Pediatrics (VIP) Network consists of >270 diverse hospitals dedicated to standardizing and improving the care of children evaluated in acute care settings.22 –26 The Better Antibiotic Selection in Children (BASiC) project conducted by the VIP Network aimed to increase the proportion of children receiving appropriate (1) empirical antibiotic therapy, (2) definitive therapy, and (3) treatment durations to at least 85% by January 1, 2022 for CAP, UTI, and SSTI.
Methods
Context
Patient Population
We included ED and hospital admission encounters of patients >60 days to 18 years old with 1 of the target infection diagnoses. Sites identified eligible encounters using the International Classification of Diseases, 10th Edition, Clinical Modification (ICD-10-CM) diagnosis codes, adapted from prior studies (Supplemental Table 3).27 ,28 Children were excluded if, upon chart review, they were found to be ill-appearing/have suspected sepsis, diagnosed with an immunocompromising condition (eg, hematologic malignancy, primary immune deficiency, etc) or cystic fibrosis, or did not receive antibiotics within the first 24 hours of the encounter. Children with other chronic complex medical conditions as defined by Feudtner, et al.29 were not explicitly excluded; chart reviewers were asked whether a patient had a complex chronic condition using diagnosis codes and groupings provided to sites (Supplemental Table 4). Patients with diagnoses or combinations of diagnoses likely to alter antibiotic therapy for an infection type (eg, chronic respiratory disease for CAP) were excluded from analysis for that infection.
Planning the Intervention
We formed an expert working group comprising specialists in hospital medicine, infectious diseases, pediatric emergency medicine, quality improvement, and antimicrobial stewardship (Supplemental Fig 5). The working group met in person and virtually from June 2019 to January 2020 to identify target measures and establish goal performance metrics for each measure. The 3 diagnoses were chosen based on their common incidence in children presenting to both community and referral acute care settings.30 During the planning phase, we created a driver diagram (Fig 1), and developed and refined the BASiC “change package” interventions (described below).
Recruitment occurred from August 3, 2020 through September 4, 2020 via announcements over applicable LISTSERVS (targeting Hospital Medicine, Infectious Diseases, and Pediatric Emergency Medicine audiences) and via a promotional webinar on July 31, 2020. Sites were required to submit a written application that described a multispecialty improvement team ideally consisting of at least 1 ED-based clinician and 1 inpatient-based clinician and demonstrate support for BASiC from senior hospital administration, including payment of the project $1000 enrollment fee. The fee was meant to cover project expenses and increase commitment to project participation among sites while minimizing the risk of posing a barrier to participation for sites with resource constraints.
Intervention
The project intervention period occurred in 2 phases (phase 1: March 2021 to June 2021 and phase 2: July 2021 to December 2021). Sites could extend data collection to March 1, 2022 because of delays in availability of the Quality Improvement Data Aggregator (QIDA). QIDA is a Web-based data collection tool wherein sites could also view real-time local and national run charts of each measure in aggregate and subdivided by condition. All sites completed and reviewed baseline assessments to facilitate implementation planning, including the Centers for Disease Control’s antimicrobial stewardship program inventory31 and the Organizational Readiness for Change Assessment.32
The BASiC “change package” comprised 6 core interventions: (1) audit with feedback33 whereby site teams entered monthly data and received graphical summaries of their site’s performance on project metrics along with project-wide, location specific (ED versus inpatient), and by-condition subcomparisons; (2) peer coaching34 in which geographically-clustered sites met with physicians experienced in mentoring and QI projects at least once per month to discuss project implementation and challenges; (3) academic detailing35 materials including posters, flyers, and buttons delivered to sites to promote local awareness of the project; (4) standardized order sets36 for each infection type; (5) educational webinars from content and implementation experts37 ; (6) a project-wide listserv for participants and leadership to answer questions and share ideas; and (7) a mobile device-based electronic clinical decision support tool.38 ,39 The mobile device-based electronic clinical decision support tool was released in phase 2 of the intervention and provided users step-wise reference for antibiotic selection based upon local sites’ antibiogram data that were submitted to project staff.
Measures
Project measures consisted of 3 outcome measures and 3 balancing measures (Table 1). Recommendations for CAP were based largely upon the Infectious Disease Society of America and Pediatric Infectious Disease Society guidelines.9 Although more recent data and systematic reviews suggest that shorter courses (ie, 5 days) may be sufficient,40 consensus was lacking during the time we developed this measure. Skin and soft tissue infection recommendations were based upon the 2014 Infectious Disease Society of America guidelines.10 Urinary tract infection recommendations were based on the 2016 AAP guidelines11 and the most recent Cochrane review for acute pyelonephritis in children.41 Although UTI is commonly treated for much shorter durations in adult women (3 days for uncomplicated cystitis),42 differentiating cystitis from pyelonephritis in children can be challenging.43 Therefore, the expert working group selected treatment duration recommendations that allowed for adequate treatment of pyelonephritis if deemed appropriate by the treating provider.
Project Measures and Goal Adherence and Outcome Rates
| Type | Name | Definition | Target |
|---|---|---|---|
| Outcome | Appropriate empirical therapy | Increase the proportion of patients started on empirical narrow-spectrum therapy based on local susceptibilities | 85% |
| Outcome | Appropriate definitive therapy | Increase the proportion of patients discharged on the narrowest spectrum definitive therapy given culture results (when present) or local susceptibilities | 90% |
| Outcome | Appropriate antibiotic duration | Increase in the proportion of patients prescribed recommended duration of antibiotic therapy based on infection type (inclusive of outpatient prescriptions): ≤10 d for CAP; ≤10 d for UTI; ≤5 d for SSTI | 85% for CAP and UTI; 90% for SSTI |
| Balancing | Revisit or readmission rate for same diagnosis within 14 d | No change in proportion of children readmitted for any diagnosis | No increase from baseline |
| Balancing | Transfer to higher level of care or ICU admission | No change in rate of transfer to ICU or higher level of care | No increase from baseline |
| Balancing | Length of hospital stay | Number of hospital days unchanged or decreased postintervention | No increase from baseline |
| Type | Name | Definition | Target |
|---|---|---|---|
| Outcome | Appropriate empirical therapy | Increase the proportion of patients started on empirical narrow-spectrum therapy based on local susceptibilities | 85% |
| Outcome | Appropriate definitive therapy | Increase the proportion of patients discharged on the narrowest spectrum definitive therapy given culture results (when present) or local susceptibilities | 90% |
| Outcome | Appropriate antibiotic duration | Increase in the proportion of patients prescribed recommended duration of antibiotic therapy based on infection type (inclusive of outpatient prescriptions): ≤10 d for CAP; ≤10 d for UTI; ≤5 d for SSTI | 85% for CAP and UTI; 90% for SSTI |
| Balancing | Revisit or readmission rate for same diagnosis within 14 d | No change in proportion of children readmitted for any diagnosis | No increase from baseline |
| Balancing | Transfer to higher level of care or ICU admission | No change in rate of transfer to ICU or higher level of care | No increase from baseline |
| Balancing | Length of hospital stay | Number of hospital days unchanged or decreased postintervention | No increase from baseline |
The expert working group also recognized that identifying optimal empirical antibiotic choices for sites would require assessment of local institutional antibiograms, whenever available. Prior studies have demonstrated that giving prescribing healthcare providers access to antibiogram data increases appropriate empirical antibiotic prescribing.44 To that end, sites were asked to submit their local antibiograms. Smaller sites that did not have pediatric-specific antibiograms submitted their hospital system’s antibiograms. Antibiograms were then used to provide site-specific empirical antibiotic recommendations for each diagnosis category within the mobile app.
Definitive therapy was defined as any narrow-spectrum antibiotic identified as susceptible on culture, or if no pathogen was identified by bacterial culture, the appropriate empirical therapy based on guidelines (eg, amoxicillin for CAP) or local antibiogram.
Study of the Intervention
Sites collected 18 months of baseline data from July 1, 2019 to December 31, 2020, 2 months of data during a washout period from January 1, 2021 to February 28, 2021, and 10 months of implementation data from March 1, 2021 to December 31, 2021. Each site was responsible for collecting its own data using a chart review tool created for the project. Participants received further guidance on appropriate data collection longitudinally via coaching calls, webinars, and the project-wide listserv. Sites were instructed to review at least 20 charts a month. Sites could prioritize diagnoses and collect unequal numbers of charts for each. Although sites could submit more than 20 charts, they were encouraged to limit submissions to 20 to avoid skewing the aggregate numbers. Sites were instructed to use the project-specific chart randomization tool for sampling if they had more than 20 eligible charts and did not wish to submit them all. Site-relative and absolute improvements in adherence were computed quarterly, with high-performing sites recognized at site-wide webinars. The optional 2-month extension of the intervention period, from January 1, 2022 to March 1, 2022, was not included in analysis because of variability in site participation. Patient race and ethnicity were collected to inform generalizability of the data and for future planned studies to investigate potential disparities in antibiotic prescribing, which have been reported for a variety of conditions in adults and children.45 –47
Analysis
We summarized patient characteristics across study periods using descriptive statistics. We used interrupted time series (ITS) analyses for each project measure in aggregate to account for pre-existing time trends.48 We also assessed goal measure adherence separately for each diagnosis. We defined statistical significance as P < .05. Analyses were conducted in R version 4.2.1 (R Core Team, Vienna, Austria).
Ethical Approval
BASiC was sponsored by the VIP Network and deemed exempt by the AAP’s Institutional Review Board. Sites obtained additional approval as deemed necessary by their local Institutional Review Boards.
Results
A total of 118 sites submitted data during the intervention period. Figure 2 shows the distribution of participating sites in BASiC by location and hospital type. Most hospitals (77 of 118 [65%]) had availability of a pediatric pharmacist; 66 of 118 (56%) stated they had a pediatric-specific antimicrobial stewardship programs. A total of 98% had pediatric ID consultant availability (25 of 118 [21%] by phone only). A total of 1412 healthcare providers and trainees participated in the project. Of these participants, 96 were pharmacists, 42 were nurses or advanced practice providers, 109 were infectious disease specialists, 784 were hospitalists, 233 were ED providers, and 107 were trainees.
Patient demographics did not meaningfully differ between study periods (Table 2). A total of 43 916 eligible encounters were reviewed (30 799 preintervention and 13 117 post).
Characteristics of Participating Hospitals and Reviewed Patient Encounters
| Total Cohort | Baseline (Jul 2019–Dec 31, 2020) | Intervention (March 1, 2021–December 31, 2021) | ||||
|---|---|---|---|---|---|---|
| N | % | N | % | N | % | |
| Overall | 43 916 | 30 799 | 70 | 13 117 | 30 | |
| Sex | ||||||
| Female | 27 900 | 64 | 19 524 | 70 | 8376 | 64 |
| Male | 16 016 | 36 | 11 275 | 37 | 4741 | 36 |
| Race and ethnicity | ||||||
| Non-Hispanic Black | 8299 | 19 | 5917 | 19 | 2382 | 18 |
| Non-Hispanic white | 17 232 | 39 | 12 056 | 39 | 5176 | 39 |
| Hispanic | 12 126 | 28 | 8299 | 27 | 3827 | 29 |
| Asian | 1243 | 3 | 835 | 3 | 408 | 3 |
| Native American/Alaska Native | 242 | 1 | 176 | 1 | 66 | 1 |
| Native Hawaiian/Pacific Islander | 247 | 1 | 161 | 1 | 86 | 1 |
| Multiple | 156 | 0 | 86 | 0 | 70 | 1 |
| Unknown | 4371 | 10 | 3269 | 11 | 1102 | 8 |
| Encounter primary diagnosis | ||||||
| CAP | 11 487 | 26 | 8213 | 27 | 3274 | 25 |
| UTI | 16 741 | 38 | 11 588 | 38 | 5153 | 39 |
| SSTI | 15 688 | 36 | 10 998 | 36 | 4690 | 36 |
| Hospital type | ||||||
| Freestanding children’s hospital | 14 447 | 33 | 10 079 | 33 | 4368 | 33 |
| Nonfreestanding children’s hospital | 14 395 | 33 | 10 358 | 34 | 4037 | 30 |
| General hospital | 5074 | 34 | 10 362 | 34 | 4712 | 36 |
| Region | ||||||
| Northeast | 6558 | 15 | 4530 | 15 | 2028 | 15 |
| South | 15 588 | 35 | 11 100 | 36 | 4488 | 34 |
| Midwest | 12 385 | 28 | 8606 | 28 | 3779 | 29 |
| West | 9385 | 21 | 6563 | 21 | 2822 | 21 |
| Age (years) | ||||||
| Median [IQR] | 5 [2–11] | 5 [2–12] | 5 [2–11] | |||
| Total Cohort | Baseline (Jul 2019–Dec 31, 2020) | Intervention (March 1, 2021–December 31, 2021) | ||||
|---|---|---|---|---|---|---|
| N | % | N | % | N | % | |
| Overall | 43 916 | 30 799 | 70 | 13 117 | 30 | |
| Sex | ||||||
| Female | 27 900 | 64 | 19 524 | 70 | 8376 | 64 |
| Male | 16 016 | 36 | 11 275 | 37 | 4741 | 36 |
| Race and ethnicity | ||||||
| Non-Hispanic Black | 8299 | 19 | 5917 | 19 | 2382 | 18 |
| Non-Hispanic white | 17 232 | 39 | 12 056 | 39 | 5176 | 39 |
| Hispanic | 12 126 | 28 | 8299 | 27 | 3827 | 29 |
| Asian | 1243 | 3 | 835 | 3 | 408 | 3 |
| Native American/Alaska Native | 242 | 1 | 176 | 1 | 66 | 1 |
| Native Hawaiian/Pacific Islander | 247 | 1 | 161 | 1 | 86 | 1 |
| Multiple | 156 | 0 | 86 | 0 | 70 | 1 |
| Unknown | 4371 | 10 | 3269 | 11 | 1102 | 8 |
| Encounter primary diagnosis | ||||||
| CAP | 11 487 | 26 | 8213 | 27 | 3274 | 25 |
| UTI | 16 741 | 38 | 11 588 | 38 | 5153 | 39 |
| SSTI | 15 688 | 36 | 10 998 | 36 | 4690 | 36 |
| Hospital type | ||||||
| Freestanding children’s hospital | 14 447 | 33 | 10 079 | 33 | 4368 | 33 |
| Nonfreestanding children’s hospital | 14 395 | 33 | 10 358 | 34 | 4037 | 30 |
| General hospital | 5074 | 34 | 10 362 | 34 | 4712 | 36 |
| Region | ||||||
| Northeast | 6558 | 15 | 4530 | 15 | 2028 | 15 |
| South | 15 588 | 35 | 11 100 | 36 | 4488 | 34 |
| Midwest | 12 385 | 28 | 8606 | 28 | 3779 | 29 |
| West | 9385 | 21 | 6563 | 21 | 2822 | 21 |
| Age (years) | ||||||
| Median [IQR] | 5 [2–11] | 5 [2–12] | 5 [2–11] | |||
Overall Adherence to Project Metrics
Median adherence [interquartile range (IQR)] to appropriate empirical, definitive, and duration of antibiotic therapy was 67% [65% to 70%], 74% [72% to 75%], and 61% [58% to 65%] at baseline, respectively, and was 72% [71% to 73%], 79% [79% to 80%], and 71% [69% to 73%], respectively, during the intervention period. Figure 3 describes the interrupted time-series analysis of adherence to project metrics. When accounting for time trends in adherence using ITS, there was a 13.1% (95% confidence interval [CI]: 0.56% to 25.58%) increase in adherence to appropriate empirical therapy at intervention start (intercept change) and no slope change. There was no intercept or slope change in adherence to definitive therapy. Adherence to appropriate antibiotic therapy duration showed no intercept change and slope increased at a rate of 1.13% (95% CI: 0.36% to 1.90%) per month above baseline adherence trends during the intervention period. Balancing metrics of escalation of care, length of stay, and revisit and readmission did not increase from baseline (Supplemental Table 5).
Subanalysis of Metric Adherence by Diagnosis
Subanalysis of adherence to project metrics stratified by primary diagnosis is summarized and depicted in Supplemental Tables 3–5 and Fig 4. For CAP, median baseline adherence [IQR] to empirical, definitive, and duration of therapy were 49% [48% to 51%], 60% [58% to 63%], and 90% [89% to 92%], respectively, and increased to 54% [52% to 56%], 72% [69% to 73%], and 93% [92% to 95%], respectively. When accounting for time trends in adherence using ITS, there were no changes in intercept or slope for all 3 metrics. For SSTI, median baseline adherence to empirical, definitive, and duration of antibiotic therapy were 94% [94% to 95%], 93% [93% to 94%], and 40% [38% to 43%], respectively, and were 96% [95% to 96%], 95% [94% to 95%], and 57% [55% to 57%], respectively during the intervention period. When accounting for time trends in adherence using ITS, there were no changes in intercept or slope for all 3 metrics. For UTI, median baseline adherence to empirical, definitive, and duration of antibiotic therapy were 52% [52% to 56%], 64% [63% to 67%], and 59% [57% to 60%], respectively, and increased to 62% [60% to 63%], 72% [71% to 74%], and 70% [68% to 71%], respectively during the intervention period. When accounting for time trends in adherence using ITS, slope increased at a rate of 0.80% (95% CI: 0.15% to 1.46%) and 0.66% (95% CI: 0.17% to 1.16%) per month above baseline for empirical and definitive therapy, respectively.
Discussion
Our QI collaborative consisting of 118 hospitals achieved overall improvements in adherence to empirical antibiotic therapy and duration of antibiotic therapy for children seen in the ED or hospital with CAP, SSTI, or UTI. Given that roughly a third of children in BASiC were seen at general hospitals with pediatric beds or wards and another third at nonfreestanding children’s hospitals, our project’s results highlight the potential for multisite QI collaboratives to effectively improve antimicrobial prescribing across diverse settings.
BASiC builds upon prior projects conducted by the VIP Network to increase appropriate antibiotic use in community-acquired pneumonia. Project iCAP was a 53-site project that aimed to improve the management of children hospitalized with community-acquired pneumonia. Conducted from 2014 to 2015, iCAP increased narrow-spectrum antibiotic use by 67% in the ED, 43% in the hospital, and 25% at discharge.22 Similar to iCAP, we found adherence to recommended empirical and definitive therapy at baseline was below goal adherence rates. However, baseline adherence rates observed in BASiC were higher than those reported in iCAP, which may explain the comparatively smaller improvements achieved in BASiC. We also found that baseline adherence to recommended treatment duration was quite high (90%). Nonetheless, given accumulating evidence suggesting that 5 days of antibiotic therapy may be sufficient to effectively treat CAP among children seen in outpatient and ED settings,49 ,50 future projects should consider focusing on reducing treatment duration.
Our findings build upon evidence from prior reports demonstrating improvements in judicious antibiotic use for SSTI and UTI through QI collaboratives. For example, a project conducted across 3 pediatric EDs and 8 urgent care centers demonstrated that adherence to appropriate antibiotic agent and duration (a bundled measure) increased from 28% to 65% for purulent SSTI and from 2% to 43% for nonpurulent SSTI.51 Kasmire et al found that implementing ED discharge order sets for SSTI and UTI at a single institution reduced inappropriate antibiotic prescriptions from 32.8% and 26.1%, respectively, to 12.5% and 13.8%, respectively at 6 months postintervention; however, antibiotic prescription duration was not assessed.52 Poole et al showed that implementing a clinical pathway in a health system’s ED and 3 urgent care centers increased use of narrow-spectrum antibiotic use for UTI from 19% to 79%.53 Our project uniquely contributes to existing literature by including hospitalized children in addition to those seen in the ED, including a diverse array of sites and multiple diagnoses to improve the ability of smaller volume centers to participate.
Although no metric achieved the goal of 85% adherence, our observed rates during the intervention period exceeded those reported in previous studies. Additionally, when accounting for time trends in adherence using ITS analysis only empirical therapy and duration of antibiotic therapy demonstrated improvements above those predicted based on baseline trends. High baseline adherence to appropriate definitive antibiotic therapy observed in SSTI (>85%) and low numbers of CAP encounters submitted during some intervention months because the coronavirus disease 2019 pandemic may account for this finding. However, subanalysis by diagnosis demonstrated increasing slope in monthly adherence rate for definitive therapy for UTI. These findings regarding definitive therapy highlight the risk of summarizing and analyzing performance metric adherence across multiple diagnoses.
Prior research has shown that diagnosis-specific quality performance metrics are difficult to assess in pediatric acute care because of small sample sizes at individual sites.54 BASiC included multiple conditions that possessed similar quality of care elements to increase the number of eligible patients at the smaller volume sites, thus allowing for more meaningful intervention assessments at these sites. Given the fact that more than two-thirds of pediatric emergency and inpatient care occurs at hospitals that are not freestanding children’s hospitals, maximizing engagement of these smaller sites is important to ensure equitable dissemination of best practices across diverse settings and communities.
This study has limitations. Sites self-selected to participate in the project; thus, our results may be biased toward greater observed effect than what would be expected of similar settings in general. Although extensive training and supplemental references were provided to sites to maximize the quality of data entered in QIDA, data quality was not formally assessed at a site level. Interventions for each project measure were implemented at different times with different amounts of fidelity among the sites, because the heterogeneity of site characteristics and the different emphasis that sites placed on each diagnosis (ie, some sites might prioritize antibiotic duration in SSTI, whereas others might prioritize improving empirical antibiotic prescribing in UTI and CAP). This flexibility was also a strength in that it helped maintain the engagement of our sites in the project. Chart selection was up to the sites and so we cannot ensure a random sample. However, sites were encouraged to use a provided case sorting tool that randomized charts within a given month and diagnosis group when sites had more charts for a given diagnosis than what they planned to enter into QIDA in a given month. Finally, we did not observe practice behavior beyond the intervention period and thus cannot assess the sustainability of changes observed beyond the project period.
Conclusions
A multisite learning collaborative to promote antimicrobial stewardship best practices improved appropriate antibiotic use across diverse settings.
Acknowledgments
We thank the teams at the 118 participating sites for their effort and commitment to this project and to the patients and families they serve.
Dr McCulloh conceptualized and designed the study, designed the data collection instruments, coordinated and supervised the collection of data, reviewed the initial analyses, and drafted the initial manuscript; Dr Kerns designed the study, designed the data collection instruments, and conducted the initial analyses; Mr Flores assisted with the analysis; Drs Cane, Feghaly, Marin, Markham, Newland, and Wang conceptualized the study; Dr Garber conceptualized and helped design the study and participated in the design of the data collection instruments; and all authors critically reviewed and revised the manuscript, approved the final manuscript as submitted, and agree to be accountable for all aspects of the work.
COMPANION PAPERS: Companions to this article can be found online at www.pediatrics.org/cgi/doi/10.1542/peds.2023-063339, www.pediatrics.org/cgi/doi/10.1542/peds.2023-063509 and www.pediatrics.org/cgi/doi/10.1542/peds.2024-065653.
FUNDING: This project was supported by the American Academy of Pediatrics Value in Inpatient Pediatrics Network. American Academy of Pediatrics staff assisted in the design and implementation of data collection tools and coordinated project activities. American Academy of Pediatrics staff had no role in data analysis, interpretation, or writing of the manuscript.
CONFLICT OF INTEREST DISCLOSURES: Drs Kerns and McCulloh share authorship rights to the software used to make the mobile decision support tool for this project, which was developed at the University of Nebraska Medical Center; Drs Kerns and McCulloh have funding from National Institutes of Health; Dr Newland has funding from Agency for Healthcare Research and Quality; and the remaining authors have no conflicts of interest relevant to this article to disclose.
Comments