Reducing Antibiotic Duration for Uncomplicated UTI in the Pediatric Emergency Department Free

Our large pediatric health system is composed of 3 freestanding children’s hospitals in an urban area within the Southeastern United States. Each hospital (2 tertiary care centers and 1 community hospital) has its own ED with a combined annual census of >200 000 patients. EDs are staffed by attending physicians trained in pediatric emergency medicine or pediatrics, and 2 of the 3 sites have advanced practice providers, residents, and fellows. The EDs share the same Epic (Epic Systems Corporation) electronic health record (EHR).

As a baseline (July 2020 to Jan 11, 2021), although 91% of children 5 to 18 years of age treated for uUTI received narrow-spectrum antibiotics based on our local antibiogram, only 13% were prescribed ≤5 days of antibiotics, and 54% received 10 days or more of antibiotics.

Our target population consisted of all patients aged 5 to 18 years who were discharged from the ED with uUTI and a prescription for antibiotics. Patient records were identified by using reports created by utilizing Structured Query Language (SQL; Supplemental Table 2) from Clarity, the reporting database for our Epic EHR. Any patients with a diagnosis of pyelonephritis or immunosuppression or a history of urological surgery, vesicoureteral reflux, or neurogenic bladder were excluded by screening past encounter diagnoses. In addition, patients with documentation of fever as a chief complaint or in a review of symptoms, back pain as a chief complaint, or a temperature >38.2° C at any time during the visit were also excluded. Thus, we had a cohort of patients discharged from the ED with the treatment of uncomplicated UTI (uUTI).

Data collected included age, sex, date of presentation, ethnicity, location of encounter, provider name, antibiotic name, dose, and duration, encounter diagnosis (as entered by the provider), urine culture results, any subsequent encounters (at the ED, urgent care, and primary care clinics of the health system) within 14 days after ED discharge, and diagnosis at any repeat encounters. The reports were manually screened to exclude any duplications, misdiagnoses, and deviations from inclusion and exclusion criteria. At the time of discharge from the ED, the treating physician prescribes the empirical antibiotic on the basis of clinical factors and without the availability of urine culture results. Availability of urine culture results was not a requirement for inclusion. Patients admitted to the hospital were not included, per the project scope. Positive urine culture result was defined as the growth of a single organism in urine culture at >50 000 colony forming units/mL.¹⁵ 

A fishbone analysis was performed to identify barriers to the prescription of short-duration antibiotics for uUTI (Fig 1). Factors included the absence of local or US national guidelines for uUTI, a lack of awareness of current evidence regarding the duration of antibiotics, fear of treatment failure, a lack of follow-up in the ED setting, and familiarity with longer prescriptions for other common bacterial infections, such as pneumonia, otitis media, streptococcal pharyngitis, etc.

The QI project was initiated with 3 comprehensive educational presentations to ED providers from January to March 2021 on current evidence, international guidelines, and results from our gap analysis, as well as the recommendation for shorter antibiotic duration. Through the course of the QI project, this message was reinforced by using posters and flyers, in-person meetings, and department-wide newsletters.

Regular feedback was provided at both the individual and group levels. Department-wide e-mails were sent to share the progress of the QI project at the group level. Biweekly chart audits were conducted, and confidential individual feedback was sent to providers by the project leader. The feedback included a report on individual performance with specific details of the patient encounters and antibiotic prescription duration, charts with group performance, and reminders of recommendations in the QI initiative.

In June 2021, an order set that was preset to select a 5-day duration of narrow-spectrum antibiotics was implemented in the EHR.

The baseline period was from July 1, 2020 to January 11, 2021, and the postimplementation period spanned from January 15, 2021 to August 15, 2022. A manual chart review was performed for patients who had received >5 days of antibiotics to search for any additional clinical factors suggestive of complicated UTI, such as the presence of costovertebral angle tenderness, nausea, or vomiting, and findings that may not have been captured electronically but may justify a longer antibiotic course. This thorough manual chart review also allowed us to provide robust feedback to individual providers to help them feel confident about changing their practice.

The primary outcome measure was the proportion of children 5 to 18 years of age with uUTI (denominator) who were prescribed ≤5 days of antibiotics (numerator) on discharge from 1 of the 3 EDs.

The secondary outcome measure was the number of antibiotic days saved. This was calculated with the assumption that, in the absence of the QI, prescribing would have continued at the mean duration calculated from the baseline period. The resultant difference of overall days of antibiotic prescription for all patients with that assumption (number of patients × mean duration) and actual total antibiotic were calculated as antibiotic days saved.

The balancing measure was designed to identify any increase in treatment failures due to a shorter duration of antibiotics compared with the baseline. This is reported as the proportion of all patients who had a culture-positive UTI at the index visit (denominator) who returned within 14 days with a second diagnosis of UTI (numerator). For return visits, we screened encounters at primary care clinics, urgent care centers, and the 3 EDs affiliated with the health system. A manual chart review of all these patients was conducted to identify treatment failure. There is no published data on the incidence of treatment failure after the antibiotic treatment of pediatric uUTI. Reported rates of persistent bacteriuria after treatment range from 0% to 23%,⁶  and the risk does not differ with the duration of antibiotics. In the absence of a known acceptable rate for the balancing measure, we chose an upper limit of 2%.

Shewhart p-charts were used to continuously evaluate the main outcome measure (short duration of antibiotics) plotted on the vertical axis against time on the horizontal axis. Each data point represents half a month of data. QI Macros (KnowWare International, Inc., Denver, CO) was used to generate the control charts with control limits set at 3-σ, and standard rules¹⁶  were used to detect special cause variation. Balancing measures are described as frequencies and proportions and were compared by using Fisher’s exact test. The level of statistical significance was set at P <.05.

Our institutional review board determined the initiative to be a nonhuman-subject project; thus, it was exempt from institutional review board review.

A total of 1292 patients who met the inclusion criteria were seen for uUTI in the duration of the project: 363 were seen in the baseline period, and 929 were seen in the post-intervention period. A total of 36% of these patients had culture-positive UTIs. The patient characteristics are given in Table 1. Cephalexin was the most prescribed antibiotic.

The proportion of patients receiving 5 days or less of antibiotics during each 15-day period increased from a baseline mean of 13% to 91% during the project (Fig 2). This short duration of antibiotics quickly increased to greater than the goal of 50% within 2 months of intervention and was sustained over a 15-month period, with a mean of 81%. There was a further shift in the center line to 91% sustained over a period of 4 months until the end of this report. The mean (SD) duration of antibiotic prescriptions improved from 8.49 (1.99) days during the baseline period to 5.67 (1.63) days during the implementation. This mean decrease in antibiotic prescription of 2.82 days per patient for the 929 patients in the postintervention period resulted in 2619 fewer antibiotic days prescribed.

During the baseline period, no patients returned to the ED with treatment failure. During the intervention period, 2 patients, a 6-year-old female and a 16-year-old sexually active male, who had confirmed uUTIs at the index visit, returned within 14 days with another UTI. This amounted to 0.6% of total culture-positive patients during this period. (P = NS). In both these cases, the identified pathogen was sensitive to the antibiotic that had been prescribed for 5 days at the initial encounter. Neither of these patients was diagnosed with pyelonephritis nor were they sick enough to require hospitalization, intravenous fluids, or intravenous antibiotics at the second visit. This revisit rate is below the previously set upper limit of 2% as the balancing measure.

A rapid, significant, and sustained increase in the proportion of short-duration antibiotic prescriptions for uUTI was achieved by using QI methodology, leading to 2619 fewer days of antibiotic prescriptions. This was not associated with a significant increase in revisits to the ED with UTI. This performance improvement and sustainment was achieved with a combination of education, targeted feedback, and EHR changes.

To our knowledge, there are no previous reports of pediatric ED-based QI for decreasing antibiotic duration specific to uUTI. A comparable QI initiative for patients between ages 3 and 18 years in outpatient clinics of a large health care organization increased the rate of short-duration antibiotics from 26.1% to 68.6% using education and EHR interventions.¹⁷  Tertiary pediatric EDs differ from pediatric outpatient clinics in that they have a much higher volume of patients, treat generally sicker patients, and lack the certainty of reliable follow-up. Previous efforts in the pediatric ED have targeted antibiotic choice for UTI but have not targeted a decrease in duration.^18–20  In another study based in pediatric urgent care clinics, Saha et al achieved a reduction in antibiotic exposure with the implementation of a protocol for urine culture follow-up and early antibiotic discontinuation for negative urine culture results.²¹  In that study, the inability to contact families to discontinue the previously prescribed antibiotics would still allow many avoidable days of antibiotics if the initial prescription was for a longer duration. In contrast, we focused on a decrease in the duration of prescribed antibiotics right at the outset. Our approach was less resource-intensive because no repeated phone calls or letters were necessary.

A national QI collaborative, Better Antibiotic Selection in Children, targeted antibiotic choice and short duration for 3 pediatric infections (UTI, pneumonia, and skin and soft tissue infections) using a multidisciplinary approach with ED, inpatient pediatrics, and infectious diseases. The results of this QI have not yet been published. The authors of a recently published large, randomized controlled trial,²²  Short Course Therapy for Urinary Tract Infections (SCOUT), compared the efficacy of standard-course (10 days) and short-course (5 days) therapy for children with UTI. This study revealed that, in children aged 2 months to 10 years, randomization to standard-course therapy had lower rates of treatment failure in comparison with short-course therapy. However, the failure rate in both treatment groups was low, and the authors suggest that short-course therapy could be considered a reasonable option for children exhibiting clinical improvement after 5 days of antimicrobial treatment. One of the criticisms²³  of SCOUT is that the authors of the study did not distinguish children with cystitis and pyelonephritis in their primary outcome analysis. This distinction is considered critical to guiding effective treatment duration. In our QI project, we excluded patients with a history of any fever, nausea, vomiting, or flank pain that may suggest upper UTI/pyelonephritis. In addition, we included only discharged patients, thus focusing on a population of truly uncomplicated UTIs. Despite a 91% rate of short-duration (5 days) antibiotics in our study, there was no significant increase in return visits for persistent or recurrent UTI within 14 days of the initial encounter. Although a return to ED does not definitively establish the safety of short-duration therapy, our findings complement those of the SCOUT trial for the management of uncomplicated UTI.

There was a rapid change of practice in our project, which was somewhat surprising and made iterative tests of change unnecessary. The likely reasons for this rapid success include a strong evidence base for the recommendations and careful selection of the target population. We believe that multiple factors, including familiarity with shorter-duration antibiotics (3 days) in adults and adolescents, the exclusion of complicated UTI, and the choice of the 5 to 18 years age group all helped in the rapid adoption of the shorter duration of therapy. ED physicians are more likely to implement change in children 5 to 18 years of age who are well-appearing and without fever, systemic symptoms, or underlying complex medical conditions. Finally, we speculate that feedback revealing a low positive culture result rate of 36% gave providers the reassurance that most of the presumed “uUTI” did not need antibiotics at all.

The second shift in centerline from 81% to 91% in April 2022 coincided with the initiation of another departmental QI project aimed to reduce unnecessary UTI screening and treatment. This second project started in the first quarter of 2022 and was not related to antibiotic duration; therefore, its impact on the current project’s outcome remains speculative. However, it is possible that the timing of this second project on a similar (UTI) topic may have helped keep awareness of the duration of antibiotics “top of mind” and thereby, indirectly helped to sustain our results.

There are several limitations to this study. First, education was the main intervention in this project, and educational interventions alone are typically considered relatively ineffective.²⁴  EHR modifications can provide level 2 reliability and hard wiring of change, but pandemic-related priority shifts caused delays in EHR changes. Our success is likely due to a combination of interventions across multiple domains of the hierarchy of reliability,²⁵  including education, dissemination of information, oversight, and individual feedback. Second, it is possible that our center’s participation in the Better Antibiotic Selection in Children QI collaborative, which also addressed short-duration antibiotics for UTI, and our own second UTI-related QI project may have influenced some of our results. These projects started 6 and 15 months, respectively, after the initiation of our QI project and were not temporally related to initial improvement. It is, however, plausible that they may have helped sustain the practice changes. In addition, interventions in an academic setting like ours may not have identical outcomes if executed in other settings, such as nonacademic settings, and may need to be modified on the basis of the analysis of local factors.

The true incidence of treatment failure or recurrence of UTI is not known,⁶  and the balancing measure, a return rate of 0.6% reported in our study, should not be considered as true recurrence. This was not the objective of this study. Our balancing measure relies on the EHR-based identification of repeat visits to an ED, urgent care, or primary care clinic within our health system, and a patient presenting for care outside these would not be identified. Additionally, a 14-day cutoff in the balancing measure is arbitrary because patients with untreated symptomatic bacteriuria may revisit after 14 days. Other published reports on antibiotic stewardship for UTI have also used a 7- to 14-day cutoff for revisits, although in those studies most patients received antibiotics for much more than 5 days.^17–20  Given our target of a 5-day antibiotic course, we felt that an additional 9 antibiotic-free days were adequate for monitoring for treatment failure. Finally, reports generated from SQL databases may contain patients who do not meet the criteria. However, all reports were manually checked, and additional manual screening of records of eligible patients was performed to eliminate mischaracterization.

In the first phase of the project, we excluded patients <5 years of age. The next phase of this project involves the systematic expansion of the lower age limit to 2 years. As of February 2023, the recommendations for shorter-duration antibiotics for uUTI are included in the teaching curriculum for new advanced practice providers (APPs). Finally, our methodology is now being replicated at our 8 urgent care centers in which an even larger number of patients can benefit from reduced antibiotic duration for uUTI.

We successfully increased the proportion of short-duration antibiotic prescriptions for the management of uUTI without a concomitant increase in return visits related to treatment failure. The improvement was seen rapidly, and the results were sustained, resulting in almost 2600 fewer antibiotic days. These strategies of provider education and feedback are now being replicated in our urgent care centers for an even bigger impact on antibiotic exposure.