Optimizing Antibiotic Treatment of Skin Infections in Pediatric Emergency and Urgent Care Centers

Children’s Healthcare of Atlanta (CHOA) is a large health care system in Atlanta with 3 EDs and 8 affiliated UC centers. With a combined 260 000 visits per year, 2 EDs are inside tertiary care hospitals and 1 is a community ED. UCs in 8 different zip codes throughout metropolitan Atlanta have a combined 198 000 visits per year. Providers include pediatric emergency medicine attending physicians (private and academic), UC physicians, pediatric emergency medicine fellows, emergency medicine residents, pediatric residents, and nurse practitioners, with 185 attending physicians practicing at these sites. In 2018, 1358 SSTI patients were seen and discharged from EDs and 1009 from UCs. A clinical practice guideline existed for outpatient purulent SSTI management but was due for revision before the start of this QI project.

We sought to include all patients seen and discharged from CHOA EDs and UCs with uncomplicated SSTIs. Inclusion criteria consisted of patients aged 0 to 18 years presenting with uncomplicated SSTIs and discharged with antibiotics from the CHOA EDs and UCs. To include a stringently defined population of uncomplicated SSTIs treated with outpatient oral antibiotics, we excluded infections frequently treated topically and more serious diagnoses that would require a broader spectrum of treatment, often in inpatient settings. Therefore, patients with erysipelas, folliculitis, felon, impetigo, lymphangitis, paronychia, perianal abscess, phlegmon, and preseptal or orbital cellulitis were excluded, in addition to patients with cephalosporin allergy. Patients with specific risk factors, such as immunocompromised status, were not excluded. Our population consisted only of discharged patients, and therefore were not considered as high risk for special considerations in antibiotic choice and duration. Data were collected using International Classification of Diseases, 10^thRevision (ICD-10), codes for purulent (L02) and nonpurulent SSTIs (L03, L08.8, L08.9), and further categorized in our data report by discharge diagnosis (Supplemental Table 3). Discharge diagnoses (as entered by the treating provider) were reviewed manually; diagnoses containing “abscess” were classified as purulent SSTIs, whereas diagnoses containing “cellulitis” were classified as nonpurulent SSTIs. Additionally, if a nonpurulent SSTI discharge diagnosis had an associated current procedural terminology (CPT) drainage procedure code, it was classified as a purulent SSTI. Based on this information, encounters were categorized as purulent and nonpurulent SSTI and filtered according to the inclusion/exclusion criteria.

The improvement team consisted of a pediatric ED fellow and attending, and a pediatric infectious diseases/antimicrobial stewardship physician and physician assistant. The SSTI clinical practice guideline revision team consisted of QI specialists, pediatric emergency medicine and urgent care physicians, and infectious diseases physicians. The improvement team identified 3 main key drivers that would influence prescribing practice change: increasing provider awareness of revised guidelines and antibiotic recommendations, obtaining provider buy-in for practice change, and making prescribing more user-friendly (Fig 1).

Clinical practice guidelines for management of SSTIs were developed and updated by an interdisciplinary team to reflect latest recommendations for antibiotic treatments of purulent and nonpurulent SSTIs.²  The preferred antibiotic choice was cephalexin for nonpurulent SSTIs and either clindamycin or trimethoprim/sulfamethoxazole for purulent SSTIs. Optimal duration of treatment was defined as 5 days for a nonpurulent SSTI and 7 days for a purulent SSTI. MRSA susceptibility rates to clindamycin in 2018 were between 82% and 91% and were 98% to trimethoprim–sulfamethoxazole. The revised guideline was released on August 1, 2019, available on an intranet Web site and within the electronic medical record (EMR) (Supplemental Fig 9).

As per our institution’s usual dissemination plan, the SSTI guidelines were shared with all ED and UC providers (including nurse practitioners) via in-person meetings, e-mail, and in a departmentwide newsletter in July and August 2019. Additionally, the antimicrobial stewardship team presented the evidence behind these revised guidelines at division meetings for ED and UC providers.

The Epic SSTI discharge order set, released August 2019, provided recommendations for antibiotics and duration of treatment of SSTIs. Categories included cellulitis (nonpurulent SSTI), abscess <1 cm (purulent SSTI, not requiring I&D), and abscess >1 cm (purulent SSTI requiring drainage). Discharge prescription order sets contained recommended SSTI antibiotic prescriptions. The antibiotic for nonpurulent SSTI was cephalexin 15 mg/kg 3 times daily for 5 days; antibiotics for purulent SSTI were clindamycin 10 mg/kg 3 times daily for 7 days or trimethoprim/sulfamethoxazole 5 mg/kg twice daily for 7 days. Frequency of electronic order set usage was not able to be evaluated. The providers were informed about the updated SSTI order set and how to access the guideline in Epic.

A MOC QI project entitled “Standardizing Treatment for SSTIs in the Pediatric ED and UC” was announced in May 2019 and began in July 2019. Fifty physicians participated (27% of the 185 prescribing attending physicians). This MOC project was extended to fellows in pediatric emergency medicine, and 2 participated. This project was a collaboration between the QI team and the infectious diseases physicians leading the CHOA Antimicrobial Stewardship Program. MOC providers were offered 3 education sessions between July and November 2019, with 2 required to obtain MOC credit. These meetings provided a forum for discussing the guideline and addressing questions and concerns from participants. MOC participants received monthly e-mails with guideline reminders and scorecards containing individual and group performance from July 2019 to December 2019. Scorecards included individual and group data for patient encounters, rates of optimal duration and choice of antibiotic, medical record numbers, and total return visits (Supplemental Table 4). After the MOC project ended, no further feedback or education was given to participants.

Data were obtained from the EMR for all eligible patients, with baseline data from January 2018 to June 2019. Our data reports use Clarity, a standard query language server database, which is connected to our EMR system (Epic) that allows for querying via Microsoft’s standard query language Server Management Studio software. Data collected included patient age, sex, date of presentation, location of encounter, prescribing provider name, ICD-10 code, discharge diagnosis (as entered by the treating provider), antibiotic prescription and duration, and CPT code for drainage procedure (10 060, 10 061). To confirm that our data report would appropriately classify purulent and nonpurulent SSTIs, chart review was conducted over 3 months (226 patients) to validate the accuracy of the ICD-10 codes, CPT codes, discharge diagnoses, and SSTI classification. Only 3 charts (1.3%) had discordant ICD-10 codes and discharge diagnoses, and all were purulent SSTIs with a discharge diagnosis indicating nonpurulent SSTI. Once CPT codes for drainage procedure were applied, we were able to achieve 100% accurate classification for purulent and nonpurulent SSTIs. All charts with ICD-10 codes for purulent SSTIs were reviewed to verify if a drainage procedure occurred; if there was a drainage procedure or if there was spontaneous drainage of the abscess, the encounter was included as a purulent SSTI. Small abscesses that did not require drainage or spontaneously drained were also included under purulent SSTIs.

Outcome measures included:

1. the proportion of patients with nonpurulent SSTI prescribed cephalexin for 5 days; and

2. the proportion of patients with purulent SSTI prescribed clindamycin or trimethoprim/sulfamethoxazole for 5 to 7 days.

Secondary outcome measures included:

1. proportion of patients prescribed cephalexin for nonpurulent SSTI;

2. proportion of patients prescribed antibiotics for 5 days for nonpurulent SSTI;

3. proportion of patients prescribed clindamycin or trimethoprim/sulfamethoxazole for purulent SSTI; and

4. proportion of patients prescribed antibiotics for 5 to 7 days for purulent SSTI.

Process measures included percentage of providers participating in the MOC project and percentage of MOC participants attending educational sessions. The balancing measure included return visits for SSTI-associated diagnosis within 10 days of the initial ED or UC visit requiring escalation of care, defined as change in antibiotics, extension of same or different antibiotic course, I&D, or admission. Return visits for non-SSTI–related diagnoses and SSTI visits that did not result in escalation of care were not included, to capture the subset of return visits that were true treatment failures related to practice change in antibiotic prescriptions.

Data from July 2019 to January 2020 were analyzed in real time, whereas data from January 2018 to June 2019 and February 2020 to March 2021 were analyzed retrospectively. P-charts were used to continuously evaluate the main outcome measure for all providers, and MOC and non-MOC participants separately. For purposes of analysis, resident and fellow physician prescriptions were included under the attending provider because trainees were directly supervised by the attending in real time. Fellow physicians acting as the attending of record and nurse practitioner prescriptions were categorized separately. Each data point represents 1 month of data. Return visits were also evaluated using a P-chart. Three σ limits were used to set the upper and lower control limits, and standard rules were used to determine special cause variation, including 8 or more values above the baseline. Proportion of optimal antibiotic choice, optimal duration, and optimal choice, plus duration, of antibiotics for purulent and for nonpurulent SSTIs were also monitored. A Fisher’s exact test was used to compare performance of MOC to non-MOC participants, with an α level of .05.

The Emory University institutional review board determined this project was not human subjects research and was thus exempt from human subjects review.

A total of 9306 SSTIs were included, with 5507 ED visits (59.2%) and 3799 UC visits (40.8%), and nearly equal visits for purulent and nonpurulent SSTIs (Table 1). During the preintervention phase (January 2018–June 2019), there were 1945 visits for purulent SSTI and 2394 for nonpurulent SSTI. During the postintervention phase (July 2019– March 2021), there were 2597 visits for purulent SSTI and 2370 visits for nonpurulent SSTI. For all visits, 50.8% of patients were female, with >50% of patients aged <10 years (Table 1). After the educational intervention in July 2019, MOC providers treated 1569 SSTIs and non-MOC providers treated 3838 SSTIs. ED and UC sites had similar compliance rates and, for purposes of analysis, were combined.

Overall improvement was noted in prescribing practices of all providers (MOC and non-MOC participants) for both purulent and nonpurulent SSTI (Fig 2 and 3). For purulent SSTI, optimal antibiotic choice, plus duration, increased from a median of 28% to 64% in the postintervention period (Fig 2), with centerline shift occurring in association with new guideline and order set. Similarly, for nonpurulent SSTI, optimal antibiotic choice, plus duration, increased from a median of 2% to 43% in the postintervention period (Fig 3).

We further evaluated provider practice stratified by antibiotic choice and antibiotic duration for purulent SSTI and nonpurulent SSTI. For purulent SSTI, optimal antibiotic choice remained stable from a median of 89% to 92%; optimal duration increased from a median of 32% to 68%. For nonpurulent SSTI, optimal antibiotic choice increased from a median of 28% to 69% and optimal duration increased from 4% to 43%.

MOC participants had similar baseline performance as non-MOC participants for both purulent and nonpurulent SSTI (Figs 4, 5, 6, 7). For purulent SSTI, optimal duration, plus antibiotic choice, increased from a baseline median of 41% to postintervention median of 84% in MOC participants; this increase was significantly more and occurred earlier than that for the non-MOC participants (Figs 4 and 5, P < .001, Table 2). For nonpurulent SSTI, optimal antibiotic choice, plus duration, increased from a baseline median of 6% to postintervention median of 68% in MOC participants; this increase was significantly more than for non-MOC participants (Figs 6 and 7, P < .001, Table 2).

All MOC providers attended at least 2 meetings (required), and 20 (40%) providers attended all 3 meetings.

Return visits requiring escalation of care remained <2% before and after intervention (Fig 8).

Antibiotic prescribing for both purulent and nonpurulent SSTIs managed in the ED and UC settings improved for both antibiotic selection and duration. MOC participants achieved significantly higher compliance with recommendations than non-MOC participants. Performance improvement was because of a combination of education, EMR modifications, and targeted feedback on improving prescribing practices. Although we did not meet our goal of 80% optimal antibiotic choice, plus duration, for all providers, we did achieve this goal within the MOC participant cohort for purulent SSTI. Although other QI projects have focused on inpatient management of SSTIs, this multisetting QI project focused on evidence-based outpatient antibiotic prescribing for pediatric SSTIs. Our results could be readily generalized to other large, tertiary pediatric care centers looking to implement QI initiatives surrounding outpatient antibiotic stewardship.

The largest improvement was seen among MOC participants. MOC projects have been shown to successfully achieve QI goals and change provider practice.^8,12,13  The directed, individual performance feedback provided to our MOC participants may have had a large impact on prescribing practices, as has also been shown in other studies.^12–15  In our MOC participant group, centerline shifts began at the beginning of the MOC project, likely corresponding to the period of monthly, individualized feedback on antibiotic prescribing practices. Although all providers received information about the revised SSTI guideline and education about its order set, monthly scorecards with individual performance may have driven larger changes in practice for MOC participants. Although the MOC participant group of 50 physicians was sizable, it was only 27% of eligible physicians. The incentive of MOC points and the superior performance of MOC participants can potentially be useful to encourage QI project participation and practice change in future QI initiatives at our institution.

Practice changes in our project were sustained for a longer time compared with another study,¹⁵  and this sustained improvement was noted for both MOC and non-MOC groups. Factors contributing to this sustained practice change were not the focus of our study. However, we did note that, although both groups maintained the improvements, the performance difference between them remained. Particularly, performance of the MOC participants did not regress to the level of the non-MOC group after the MOC project ended. Although this is encouraging, we realize there is room for additional improvements.

Among all providers, return visits requiring escalation in care did not increase after intervention (Fig 8). This suggests the shorter duration and narrower spectrum of antibiotics did not negatively impact outcomes, supporting similar findings in previous studies.^2,9,16  This contrasts 1 study, finding purulent SSTIs treated 10 days or longer after I&D had decreased recurrence after 6 weeks.¹⁷  This retrospective review had a small sample size of patients completing 6 or 7 days of therapy (majority >7 or >5).¹⁷  With most purulent SSTI treatment lasting 7 days in our project and a small and consistent percentage of return visits, albeit a shorter follow-up timeline, further research would be needed to examine the difference in short- and long-term recurrence for 7 days versus 10 days of treatment.

Compliance with antibiotic recommendations was higher for purulent SSTI compared with nonpurulent SSTI. Purulent SSTI-noncompliant prescriptions were most commonly for longer duration (10 days), but recommended antibiotic choice was prescribed >90% of the time. For nonpurulent SSTI prescriptions, most noncompliance surrounded duration of cephalexin prescriptions (commonly 7 or 10 days in duration instead of the recommended 5 days) and, less frequently, prescribing clindamycin instead of cephalexin. There was a decrease in compliance for nonpurulent SSTI prescriptions in December and January 2021, but this trend improved in February and March 2021. Although we do not have an explanation for this decrease, we were reassured by the improvement in subsequent months, and plan to continue to monitor performance of all providers and encourage use of the SSTI guidelines.

Although we observed overall improvement in prescribing practices, we did not reach our goals in some metrics because of several possible factors. First, after the MOC project ended in January 2020, we had planned to expand the project to all providers with periodic group and individual feedback. However, this plan had to be postponed because of the disruptions caused by the coronavirus disease 2019 pandemic. Second, our providers expressed ongoing hesitation about prescribing antibiotics for a shorter duration than they were accustomed to, in addition to not covering for MRSA when treating nonpurulent SSTI, despite stable return visits. This fear of treatment failure or hesitation to adopt new guidelines has been cited in the literature.¹⁸  Third, although we strived to exclude complicated cases of SSTIs, we were not always able to identify patients who were initially treated outside of our health care system and presented only after failed antibiotic therapy. Previous treatment failure is likely to steer antibiotic choice and duration in the direction of broader spectrum and longer duration therapy. Finally, regarding return visits, only revisits in the CHOA system were included. However, balancing measures were defined and tracked the same way before and after the intervention.

We demonstrate improved optimal choice and reduced duration of antibiotic treatment of outpatient management of both purulent and nonpurulent SSTIs. Despite use of narrower spectrum antibiotics and/or shorter duration of therapy, return visits did not increase. MOC project participation was associated with greater improvement in practice and was sustained after intervention period. Further studies are needed to understand the impact of MOC projects on practice change, as well as on sustaining performance.

CHOA

Children’s Healthcare of Atlanta

CPT

current procedural terminology

ED

emergency department

EMR

electronic medical record

I&D

incision and drainage

ICD-10

International Classification of Diseases, 10th Revision

MOC

maintenance of certification

MRSA

methicillin-resistant Staphylococcus aureus

QI

quality improvement

SSTI

skin and soft tissue infection

UC

urgent care