BACKGROUND:

Prolonged antibiotic therapy may be associated with increased adverse events and antibiotic resistance. We deployed an intervention in the electronic health record (EHR) to reduce antibiotic duration for pediatric outpatients.

METHODS:

A preintervention and postintervention interrupted time series analysis of antibiotic duration for 7 antibiotics was performed for patients discharged from the ED and clinics of a children’s hospital network from 2012 to 2018. In February 2015, clickable 5- and 7-day duration option buttons were deployed in the EHR for clindamycin, cephalexin, ciprofloxacin and levofloxacin, trimethoprim-sulfamethoxazole, amoxicillin, and cefdinir, with an additional 10-day option for the latter 2. Prescribers were able to enter a free-text duration. The option buttons were not announced, and were not linked to a specific diagnosis or quality improvement initiative. The primary outcome was proportion of prescriptions per month with duration of 10 days. Balancing secondary outcomes were reorders of the same agent, return to clinic, and inpatient admissions within 30 days.

RESULTS:

There were 54 315 prescriptions for the 7 antibiotics associated with 39 894 patients, 18 683 clinic visits, and 35 632 ED visits. Overall, a −5.1% (95% confidence interval [CI], −8.3% to −2.0%) change in the proportion of prescriptions with a 10-day duration was attributable to the intervention, with larger effects noted for clindamycin (−20.8% [95% CI, −26.9% to −14.7%]) and cephalexin (−9.9% [95% CI, −14.3% to −5.4%]). There was no increase in the reorders of the same agent, return clinical encounters, or inpatient admissions within 30 days.

CONCLUSIONS:

A simple intervention in the EHR can safely reduce duration of antibiotic therapy.

What’s Known on This Subject:

Prolonged antibiotic exposure is associated with adverse events and resistance. In pediatric ambulatory care settings, interventions in the electronic health record can be important tools for antimicrobial stewardship, but prescribers may find alerts and prompts to be burdensome and time-consuming.

What this Study Adds:

In this preintervention and postintervention interrupted time series analysis, simply displaying 5- and 7-day duration option buttons in the electronic health record safely decreased the proportion of antibiotics with a 10-day duration by 5.1%, with a 20.8% reduction for clindamycin.

The Centers for Disease Control and Prevention and the Center for Medicare and Medicaid Services has recognized antimicrobial stewardship in ambulatory care settings as an important public health priority to combat antimicrobial resistance.1,2  Approximately 30% to 50% of the 60 million prescriptions per year in patients aged <20 years are unnecessary, including the treatment of syndromes in which antibiotics are not indicated and prolonged duration of therapy.3  Shorter duration of therapy may prevent the emergence of antibiotic resistance, lessen disruption of the microbiome, and decrease adverse events.4 

Clinical decision support (CDS) tools provide timely information to help inform decisions about a patient’s care.5  CDS tools at the time of prescribing are recommended for incorporation into the electronic health record (EHR) to enhance antimicrobial stewardship.6  However, CDS tools may be expensive and require extensive information technology resources. Furthermore, an excessive number of alerts can lead to alert fatigue and undermine the system’s effectiveness.7 

To decrease antibiotic use in the pediatric outpatient setting, defined as clinic visits (primary, urgent, and specialty care) and emergency department (ED) visits without admissions, we created short-duration buttons for antibiotic prescriptions in the EHR. The purpose of this study was to examine the effect of this simple intervention on prolonged duration of therapy. Furthermore, we evaluated the safety of the interventions by measuring antibiotic reorders and return clinical encounters.

The intervention was implemented at the Ann & Robert H. Lurie Children’s Hospital (LCH) of Chicago, a 364-bed, quaternary-care pediatric hospital located in Chicago, Illinois. The staff include >1600 physicians and allied health professionals in 70 pediatric specialties. In 2019, ∼56 000 visits were made to 1 ED, and 700 000 visits were made to its 50 primary and specialty clinics (including 11 000 urgent care visits) at 14 centers throughout the region.

All sites in the LCH system use the Epic EHR software application for facilitating and documenting patient care, including the ordering of medications. Before February 2015, providers prescribing antibiotics through the application typed in the number of doses or the number of days of therapy as free text as part of the ordering workflow. In February 2015, a duration option button (DOB) was deployed for all ED and clinic setting orders of the following agents: amoxicillin, cefdinir, cephalexin, clindamycin, fluoroquinolones (ciprofloxacin and levofloxacin), and trimethoprim-sulfamethoxazole. The DOB consisted of clickable buttons for 5 days and 7 days located to the right of the free-text box on the line for duration. For 2 agents, amoxicillin and cefdinir, a 10-day option was also included because these drugs are commonly used to treat acute pharyngitis and/or acute otitis media for that duration.8,9  DOBs with longer durations beside these were deliberately not provided. The display did not suggest a preference, and prescribers remained able to enter a free-text duration. The DOBs appeared suddenly, without announcement, accompanying education, or a quality improvement initiative. The total development time for the intervention was 10 hours, including design of the DOBs, testing in the Epic proof of concept environment, and deployment.

All ED and clinic (primary, urgent, and specialty care) orders of the 7 agents from the years 2012–2018 were extracted from the Epic reporting database, along with details of LCH visits for the patients associated with these orders. ED orders were written for prescriptions given to patients after discharge from the ED. Clinic orders were defined as those associated with encounter types of office visit or office procedure. Descriptive statistics related to outcomes (ie, the number of orders and prescribed duration of therapy) and balancing measures to identify potential treatment failures within 30 days of the index prescription were generated by period and care setting. Patients could be included more than once if >1 study antibiotic was ordered. The balancing metrics were defined as reorders of the same agent to approximate extension or restart of therapy, repeat visit to the clinic to quantitate the need for reevaluation, and inpatient admission to estimate escalation in care. Duration in days was determined from the total number of doses and dosing frequency. If fractions of a day were prescribed, the duration was rounded down to the nearest integer.

Interrupted time series (ITS) analyses were conducted by using the monthly proportion of prescriptions with 10-day durations and the monthly proportion for each balancing measure. ITS using the monthly mean values of prescribed duration of therapy was also performed. For the ITS analyses, records with a prescribed duration of therapy of <4 or >21 days were excluded. Because there are no known standards of care for treatment of infections in <4 days in pediatric patients, these prescription durations were considered likely to be continuations of previous courses. Prescription durations of >21 days were excluded on the basis of the likelihood of them being for daily prophylaxis for chronic conditions or for treatment of complex infections, such as osteomyelitis, that require long durations. We performed a manual review of durations that were 6 and 8 days to confirm the orders were not for 7 days.

All data set preparation, descriptive statistics, and ITS analyses were conducted by using SAS Enterprise Guide version 7.1 (SAS Institute, Inc, Cary, NC). Data were adjusted by SAS procedure X11 to remove seasonal and calendar influences before subjected to ITS modeling. Segmented autoregression models were then fitted by using ordinary least-squares regression analysis. When the Durbin-Watson test statistic was significant, we used a maximum likelihood method with the backstep option to correct for autocorrelation within the autoregression models. The white noise test was used to support model adequacy conclusions. Statistical significance was defined as P < .05.

Over the 7-year period evaluated, there were 54 315 prescriptions for the study antibiotics associated with 39 894 patients, 18 683 clinic visits, and 35 632 ED visits. The overall number of patients was less than the sum of the patients in the preintervention and postintervention periods because 3296 patients received prescriptions in both the preintervention and postintervention periods. Compared with the preintervention period, the percentage of orders with durations that appeared in the option button increased: 5-day durations increased by 1.05% (P < .001) and 7-day durations by 13.09% (P < .001). The largest percentage change (−16.26%; P < .001) occurred in 10-day durations, decreasing from 70.35% preintervention to 54.09% postintervention (Table 1).

TABLE 1

Frequency of Prescribed Antibiotic Duration for ED and Clinic Orders of Selected Antibiotics, January 2012 to January 2015 (Preintervention) and February 2015 to December 2018 (Postintervention)

Prescribed Duration, dPreintervention, No. (%)Postintervention, No. (%)Percentage Difference, % (95% CI)P
<4 234 (1.04) 378 (1.19) 0.16 (−1.55 to 1.85) .09 
32 (0.14) 38 (0.12) −0.02 (−1.72 to 1.68) .49 
497 (2.20) 1031 (3.25) 1.05 (−0.63 to 2.73) <.001 
56 (0.25) 188 (0.59) 0.34 (−1.37 to 2.05) <.001 
3296 (14.58) 8776 (27.67) 13.09 (11.56 to 14.62) <.001 
23 (0.10) 20 (0.06) −0.04 (−1.72 to 1.64) .11 
429 (1.90) 667 (2.10) 0.20 (−1.49 to 1.89) .09 
10 15 899 (70.35) 17 154 (54.09) −16.26 (−17.29 to −15.23) <.001 
11–21 1111 (4.92) 1238 (3.90) −1.01 (−2.69 to 0.65) <.001 
>21 1023 (4.53) 2225 (7.02) 2.49 2.49 (0.83 to 4.15) <.001 
Total 22 600 (100) 31 715 (100) — — 
Prescribed Duration, dPreintervention, No. (%)Postintervention, No. (%)Percentage Difference, % (95% CI)P
<4 234 (1.04) 378 (1.19) 0.16 (−1.55 to 1.85) .09 
32 (0.14) 38 (0.12) −0.02 (−1.72 to 1.68) .49 
497 (2.20) 1031 (3.25) 1.05 (−0.63 to 2.73) <.001 
56 (0.25) 188 (0.59) 0.34 (−1.37 to 2.05) <.001 
3296 (14.58) 8776 (27.67) 13.09 (11.56 to 14.62) <.001 
23 (0.10) 20 (0.06) −0.04 (−1.72 to 1.64) .11 
429 (1.90) 667 (2.10) 0.20 (−1.49 to 1.89) .09 
10 15 899 (70.35) 17 154 (54.09) −16.26 (−17.29 to −15.23) <.001 
11–21 1111 (4.92) 1238 (3.90) −1.01 (−2.69 to 0.65) <.001 
>21 1023 (4.53) 2225 (7.02) 2.49 2.49 (0.83 to 4.15) <.001 
Total 22 600 (100) 31 715 (100) — — 

—, not applicable.

In ́Table 2, we provide a summary of the ITS analyses of the monthly proportion of 10-day prescriptions, stratified by prescription setting and agent. The number of orders per month was stable in the preintervention and postintervention periods, with a slight increase in monthly median values overall, from 535 to 599 orders per month. Agent-specific medians of orders per month were also similar. The median number of orders in the ED remained similar (386 vs 384 orders per month) but increased by 35%, from 148 to 200 orders per month, in the clinic setting. The ITS results encompassing all study agents are displayed graphically in Fig 1. For the combined data set, and in the ED, there was a decreasing slope in the proportion of 10-day durations, suggesting a trend that preceded the intervention. However, an immediate level change was associated with the intervention, with changes in 10-day durations of −9.3% (95% confidence interval [CI], −12.5% to −6.2%), −5.4% (95% CI, −10.2% to −0.5%), and −5.1% (95% CI, −8.3% to −2.0%) in the clinic, ED, and combined settings, respectively. The combined-setting postintervention slope change was slightly positive, indicative of a less steep downward trend that continued after the intervention. When examining the impact on the overall change in mean duration for all prescriptions, there was a significant level change of −0.246 (95% CI, −0.662 to 0.170) days overall from the intercept (baseline) of 9.81 days (Fig 2). The cumulative intervention attributable effect was a reduction of 4969 days of therapy for the study antibiotics over the 47-month postintervention period.

TABLE 2

Summary Statistics for the Monthly Number of Orders and Seasonally Adjusted ITS Analyses of the Proportion of Prescriptions With 10-Day Duration of Therapy Ordered, by Care Setting and Selected Antibiotics, During January 2012 to January 2015 (Preintervention) and February 2015 to December 2018 (Postintervention)

Setting and AgentMedian Monthly No. Orders (IQR)ITS Estimate, % (95% CI) of 10-d Duration Proportion
PreinterventionPostinterventionInterceptPreintervention SlopeIntervention Level ChangePostintervention Slope Change
Setting       
 ED and clinic 535 (456 to 661) 599 (481 to 715) 79.3 (76.5 to 82.0)* −0.3 (−0.5 to −0.2)* −5.1 (−8.3 to −2.0)* 0.2 (0.0 to 0.3)* 
 ED only 386 (326 to 475) 384 (335 to 504) 84.9 (80.3 to 89.5)* −0.4 (−0.6 to −0.2)* −5.4 (−10.2 to −0.5)* 0.1 (−0.2 to 0.4) 
 Clinic only 148 (118 to 163) 200 (167 to 235) 54.6 (52.3 to 56.9)* 0.2 (0.0 to 0.3)* −9.3 (−12.5 to −6.2)* −0.1 (−0.2 to 0.1) 
Agent       
 Amoxicillin 318 (219 to 405) 337 (236 to 471) 91.4 (88.8 to 94.0)* −0.2 (−0.4 to −0.1)* −3.4 (−6.7 to −0.2)* 0.0 (−0.1 to 0.2) 
 Cefdinir 87 (75 to 103) 74 (62 to 83) 75.3 (72.0 to 78.7)* −0.3 (−0.4 to −0.1)* −4.9 (−9.3 to −0.6)* 0.2 (−0.0 to 0.4) 
 Cephalexin 52 (46 to 56) 68 (58 to 77) 48.5 (45.3 to 51.8)* −0.3 (−0.4 to −0.1)* −9.9 (−14.3 to −5.4)* 0.1 (−0.0 to 0.3) 
 Clindamycin 71 (60 to 76) 61 (51 to 69) 67.9 (63.2 to 72.5)* −0.3 (−0.6 to −0.1)* −20.8 (−26.9 to −14.7)* 0.1 (−0.2 to 0.3) 
 Fluoroquinolone 15 (12 to 18) 25 (22 to 30) — — — — 
 Trimethoprim-sulfamethoxazole 17 (15 to 19) 18 (15 to 20) — — — — 
Setting and AgentMedian Monthly No. Orders (IQR)ITS Estimate, % (95% CI) of 10-d Duration Proportion
PreinterventionPostinterventionInterceptPreintervention SlopeIntervention Level ChangePostintervention Slope Change
Setting       
 ED and clinic 535 (456 to 661) 599 (481 to 715) 79.3 (76.5 to 82.0)* −0.3 (−0.5 to −0.2)* −5.1 (−8.3 to −2.0)* 0.2 (0.0 to 0.3)* 
 ED only 386 (326 to 475) 384 (335 to 504) 84.9 (80.3 to 89.5)* −0.4 (−0.6 to −0.2)* −5.4 (−10.2 to −0.5)* 0.1 (−0.2 to 0.4) 
 Clinic only 148 (118 to 163) 200 (167 to 235) 54.6 (52.3 to 56.9)* 0.2 (0.0 to 0.3)* −9.3 (−12.5 to −6.2)* −0.1 (−0.2 to 0.1) 
Agent       
 Amoxicillin 318 (219 to 405) 337 (236 to 471) 91.4 (88.8 to 94.0)* −0.2 (−0.4 to −0.1)* −3.4 (−6.7 to −0.2)* 0.0 (−0.1 to 0.2) 
 Cefdinir 87 (75 to 103) 74 (62 to 83) 75.3 (72.0 to 78.7)* −0.3 (−0.4 to −0.1)* −4.9 (−9.3 to −0.6)* 0.2 (−0.0 to 0.4) 
 Cephalexin 52 (46 to 56) 68 (58 to 77) 48.5 (45.3 to 51.8)* −0.3 (−0.4 to −0.1)* −9.9 (−14.3 to −5.4)* 0.1 (−0.0 to 0.3) 
 Clindamycin 71 (60 to 76) 61 (51 to 69) 67.9 (63.2 to 72.5)* −0.3 (−0.6 to −0.1)* −20.8 (−26.9 to −14.7)* 0.1 (−0.2 to 0.3) 
 Fluoroquinolone 15 (12 to 18) 25 (22 to 30) — — — — 
 Trimethoprim-sulfamethoxazole 17 (15 to 19) 18 (15 to 20) — — — — 

IQR, interquartile range; —, not applicable.

*

P < .05.

FIGURE 1

Seasonally adjusted ITS models of the proportion of 10-day prescribed durations of therapy for selected antibiotics. A, ED and clinic orders. B, ED orders alone. C, Clinic orders alone. The dot represents the observed mean, the solid line represents the seasonally adjusted regression model trend, and the dashed line represents the deseasonalized trend.

FIGURE 1

Seasonally adjusted ITS models of the proportion of 10-day prescribed durations of therapy for selected antibiotics. A, ED and clinic orders. B, ED orders alone. C, Clinic orders alone. The dot represents the observed mean, the solid line represents the seasonally adjusted regression model trend, and the dashed line represents the deseasonalized trend.

Close modal
FIGURE 2

Seasonally adjusted ITS models of prescribed duration of therapy in days for ED and clinic orders of selected antibiotics. The dot represents the observed mean, the solid line represents the seasonally adjusted regression model trend, and the dashed line represents the deseasonalized trend.

FIGURE 2

Seasonally adjusted ITS models of prescribed duration of therapy in days for ED and clinic orders of selected antibiotics. The dot represents the observed mean, the solid line represents the seasonally adjusted regression model trend, and the dashed line represents the deseasonalized trend.

Close modal

Overall, the level changes resulted in fewer prescribed 10-day durations: clindamycin −20.8% (95% CI, −26.9% to −14.7%), cephalexin −9.9% (95% CI, −14.3% to −5.4%), cefdinir −4.9% (95% CI, −9.3% to −0.6%), and amoxicillin −3.4% (95% CI, −6.7% to −0.2%). None of them experienced a significant slope change in the postintervention period. The fluoroquinolone and trimethoprim-sulfamethoxazole prescription durations were analyzed by χ2 analysis and not by ITS because the median number of orders per month for both agents was <25 in both the preintervention and postintervention periods. The proportion of fluoroquinolone prescriptions per month prescribed for 10 days was 21.5% before the intervention and 20.0% after the intervention (P = .45). The proportion of trimethoprim-sulfamethoxazole prescriptions prescribed for 10 days was 30.1% before the intervention and 19.7% after the intervention (P < .001). ITS analyses of the 3 balancing measures provided evidence that there were no increases in reorders of the same antibiotic, inpatient admissions, or return clinic visits within 30 days (Table 3). Return clinic visits had a level change of −1.3% (95% CI, −2.6% to −0.0%) attributable to intervention.

TABLE 3

ITS Analyses of Antibiotic Reorder, Inpatient Admission, and Return Clinic Visit Within 30 Days, January 2012 to January 2015 (Preintervention) and February 2015 to December 2018 (Postintervention)

Event Occurrence Within 30 dMonthly No. Events, Median (IQR)ITS Estimate, % (95% CI)
PreinterventionPostinterventionInterceptPreintervention SlopeIntervention Level ChangePostintervention Slope Change
Same antibiotic reordered 0 (0 to 6) 0 (0 to 8) 0.6 (0.4 to 0.9)* −0.0 (−0.0 to 0.0) 0.2 (−0.1 to 0.6) 0.0 (−0.0 to 0.0) 
Inpatient admission 6 (4 to 6) 5 (3 to 7) 0.9 (0.6 to 1.2)* 0.0 (−0.0 to 0.0) 0.1 (−0.2 to 0.4) −0.0 (−0.0 to 0.0) 
Clinic visit 97 (85 to 112) 117 (102 to 128) 15.9 (15.0 to 16.9)* 0.1 (0.1 to 0.2)* −1.3 (−2.6 to −0.0)* −0.1 (−0.2 to −0.1)* 
Event Occurrence Within 30 dMonthly No. Events, Median (IQR)ITS Estimate, % (95% CI)
PreinterventionPostinterventionInterceptPreintervention SlopeIntervention Level ChangePostintervention Slope Change
Same antibiotic reordered 0 (0 to 6) 0 (0 to 8) 0.6 (0.4 to 0.9)* −0.0 (−0.0 to 0.0) 0.2 (−0.1 to 0.6) 0.0 (−0.0 to 0.0) 
Inpatient admission 6 (4 to 6) 5 (3 to 7) 0.9 (0.6 to 1.2)* 0.0 (−0.0 to 0.0) 0.1 (−0.2 to 0.4) −0.0 (−0.0 to 0.0) 
Clinic visit 97 (85 to 112) 117 (102 to 128) 15.9 (15.0 to 16.9)* 0.1 (0.1 to 0.2)* −1.3 (−2.6 to −0.0)* −0.1 (−0.2 to −0.1)* 
*

P < .05.

We demonstrated that a simple, low-cost tool decreased the aggregate proportion of 10-day prescription durations for 7 commonly prescribed antibiotics in the pediatric outpatient setting. Even with decreasing trends in antibiotic use for multiple agents in the preintervention and postintervention period, ITS analysis confirmed the immediate effect of DOB on decreasing the proportion of 10-day durations by 5.1%, with 9.3% reductions seen in the clinic setting and 5.4% seen in the ED. Furthermore, the mean duration of prescriptions decreased by 0.251 days for all study antibiotics in aggregate, and 4969 antibiotic days were estimated to have been avoided. We surmise that the 2.5% increase in durations of >21 days was a coincident secular trend because it seems unlikely that DOBs would cause clinicians to increase prescriptions with durations of >21 days but prescribe fewer 10-day durations. The increase in the median number of prescriptions per month in the postintervention period reflected an increase in clinic visits because the number of prescriptions per patient and per visit were essentially unchanged. Our intervention did not result in an increase in adverse events, measured by return to clinic, hospital admission, or antibiotic reorders. These outcomes were accomplished through a one-time intervention in the EHR, requiring only 10 hours of effort to implement.

Our intervention revealed that clinicians were more likely to choose a shorter duration of therapy when visually presented with it as an option, even without an indicated preference or accompanying stewardship messaging. Marked variation exists in the duration of antibiotic therapy for common infections, such as urinary tract infections, community-acquired pneumonia, and skin and soft tissue infections.1012  Drivers for longer duration of therapy may include patient-level differences in infection severity or concern for relapse by treating pediatricians.13  However, for many pediatric conditions, a broad range of prescribing durations are recommended by professional guidelines because of a lack of convincing data on a minimum effective duration of therapy.14,15  Even when such data exist, clinicians may default to longer durations as a force of habit or may prescribe medications longer than recommended, believing that the marginal benefits of additional therapy outweigh the risks.

Even modest decreases in duration of therapy could be beneficial in reducing adverse events and antibiotic resistance. In a multicenter cohort study of hospitalized patients with pneumonia, each excess day of antibiotic therapy over the recommended clinical care guidelines (median excess of 2 days) was associated with a 5% increase in the odds of antibiotic-associated adverse events.12  In a case control study of ampicillin-resistant urinary tract infections, previous receipt of amoxicillin for ≥7 days, but not <7 days, was associated with ampicillin-resistant urinary tract infections. For infants in the NICU, each additional day of antibiotic therapy is associated with less diversity of anaerobic bacteria in stool.16  There are >250 million antibiotic prescriptions per year in the United States in the community setting, with 60 million in patients aged <20 years.17  Even minimal decreases in antibiotic duration may reduce millions of days of antibiotic exposure.

Our intervention did not require alerts, messages, or warnings and did not prevent clinicians from prescribing durations of their preference. Even when conveying important information, prescribers may find an excessive number of alerts to be burdensome and time-consuming.18  Excessive alerts in the EHR can lead to alert fatigue, whereby subsequent alerts (including potentially serious ones) are disregarded, and opportunities for stewardship are missed.19  Prescribers may question the safety of automated suggestions for durations of therapy.7  Visually presenting, but not requiring (or even recommending), duration options nudged antibiotic decision-making toward shorter treatment. This paradigm of subtle reminders was successfully used to decrease inappropriate antibiotic prescribing for acute respiratory tract infections by displaying poster-sized commitment letters to antimicrobial stewardship in examination rooms.20 

We found variation by both antibiotic and prescribing location. When compared with the other antibiotics, cephalexin and clindamycin had larger decreases in antibiotic duration attributable to the DOBs. We surmise that this is because these agents are commonly used to treat skin and soft tissue infections for which optimal duration of therapy is less defined.10  A multifaceted quality improvement intervention has been revealed to decrease duration of therapy for skin and soft tissue infections in children.21  Decreases in the proportion of 10-day durations in the outpatient setting were greater than in the ED. We hypothesize that ED physicians may choose a longer duration of therapy because of higher acuity or for higher severity of illness, even for patients being discharged. Our findings suggest that DOBs could reduce the duration of antibiotic course in urgent care centers that have high rates of antibiotic prescribing.22 

There were several limitations with our study. We did not examine the association between antibiotic durations and specific diagnoses, appropriateness of use, or patient-level factors, such as severity of illness and chronic medical conditions. We are unable to retrospectively determine which durations were entered via duration buttons versus free text for a given duration. As the changes in percentages in 7- and 10-day durations were substantially larger than 6-, 8-, and 9-day durations, we inferred this was via direct use of the button or influenced by it. Potential adverse events were measured monthly in aggregate and not at the patient-level. It is possible that patients who had infection relapse sought care at outside hospitals. Patients may have received alternative agents after treatment failure, which we did not measure.

Our study revealed that an inexpensive, nearly invisible antimicrobial stewardship intervention can reduce antibiotic-prescribing duration. This tool could be used easily in EHRs, with minimal effort and cost. DOBs could be updated with shorter options as additional evidence emerges and clinical guidelines are updated. In future studies, researchers should investigate the impact of short-duration DOBs on specific diagnoses or with outlier-prescribing clinicians. Similarly, this tool could be used in other clinical settings, such as hospitalizations or telemedicine visits. Antimicrobial stewardship requires a range of strategies, and even small decreases in antibiotic use can contribute to reducing the burden of antimicrobial resistance.

We thank George Lales for his assistance in extracting data for the analyses.

Dr S. Patel conceptualized and designed the study and revised the manuscript; Drs R. Patel, Healy, and Scardina conceptualized and designed the study; Mr Jones and Drs Sun performed the data analysis and drafted the manuscript; Dr Fricchione conceptualized and designed the study, performed the data analysis, and drafted the manuscript; and all authors reviewed the manuscript, agree to be accountable for all aspects of the work, and approved the final manuscript as submitted.

FUNDING: No external funding.

     
  • CDS

    clinical decision support

  •  
  • CI

    confidence interval

  •  
  • DOB

    duration option button

  •  
  • ED

    emergency department

  •  
  • EHR

    electronic health record

  •  
  • ITS

    interrupted time series

  •  
  • LCH

    Lurie Children’s Hospital

1
Centers for Disease Control and Prevention
.
Core elements of outpatient antibiotic stewardship
.
2
The Joint Commission
.
Antimicrobial stewardship
.
2020
.
3
The Pew Charitable Trusts
.
Antibiotic use in outpatient settings
.
4
Centers for Disease Control and Prevention
.
Biggest threats and data: 2019 AR threats report
.
2020
.
Available at: https://www.cdc.gov/drugresistance/biggest-threats.html. Accessed April 28, 2020
5
Agency for Healthcare Research and Quality
.
Clinical decision support
.
6
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
7
Forrest
GN
,
Van Schooneveld
TC
,
Kullar
R
,
Schulz
LT
,
Duong
P
,
Postelnick
M
.
Use of electronic health records and clinical decision support systems for antimicrobial stewardship
.
Clin Infect Dis
.
2014
;
59
(
suppl 3
):
S122
S133
8
Committee on Infectious Diseases American Academy of Pediatrics
.
Red Book
, 31st ed.
Itasca, IL
:
American Academy of Pediatrics
;
2018
9
Lieberthal
AS
,
Carroll
AE
,
Chonmaitree
T
, et al
.
The diagnosis and management of acute otitis media. [published correction appears in Pediatrics. 2013;131(3):e964-e999]
.
Pediatrics
.
2013
;
131
(
3
).
10
Stevens
DL
,
Bisno
AL
,
Chambers
HF
, et al
.
Practice guidelines for the diagnosis and management of skin and soft tissue infections: 2014 update by the Infectious Diseases Society of America
.
Clin Infect Dis
.
2014
;
59
(
2
):
147
159
11
Lewis-de Los Angeles
WW
,
Thurm
C
,
Hersh
AL
, et al
.
Trends in intravenous antibiotic duration for urinary tract infections in young infants
.
Pediatrics
.
2017
;
140
(
6
):
e20171021
12
Vaughn
VM
,
Flanders
SA
,
Snyder
A
, et al
.
Excess antibiotic treatment duration and adverse events in patients hospitalized with pneumonia: a multihospital cohort study
.
Ann Intern Med
.
2019
;
171
(
3
):
153
163
13
Pakyz
AL
,
Moczygemba
LR
,
VanderWielen
LM
,
Edmond
MB
,
Stevens
MP
,
Kuzel
AJ
.
Facilitators and barriers to implementing antimicrobial stewardship strategies: results from a qualitative study
.
Am J Infect Control
.
2014
;
42
(
suppl 10
):
S257
S263
14
Roberts
KB
;
Subcommittee on Urinary Tract Infection, Steering Committee on Quality Improvement and Management
.
Urinary tract infection: clinical practice guideline for the diagnosis and management of the initial UTI in febrile infants and children 2 to 24 months
.
Pediatrics
.
2011
;
128
(
3
):
595
610
15
Bradley
JS
,
Byington
CL
,
Shah
SS
, et al;
Pediatric Infectious Diseases Society and the Infectious Diseases Society of America
.
The management of community-acquired pneumonia in infants and children older than 3 months of age: clinical practice guidelines by the Pediatric Infectious Diseases Society and the Infectious Diseases Society of America
.
Clin Infect Dis
.
2011
;
53
(
7
):
e25
e76
16
Rooney
AM
,
Timberlake
K
,
Brown
KA
, et al
.
Each additional day of antibiotics is associated with lower gut anaerobes in neonatal intensive care unit patients
.
Clin Infect Dis
.
2020
;
70
(
12
):
2553
2560
17
Centers for Disease Control and Prevention
.
Antibiotic use in outpatient settings
.
18
Eslami
S
,
Abu-Hanna
A
,
de Keizer
NF
.
Evaluation of outpatient computerized physician medication order entry systems: a systematic review
.
J Am Med Inform Assoc
.
2007
;
14
(
4
):
400
406
19
Koppel
R
,
Metlay
JP
,
Cohen
A
, et al
.
Role of computerized physician order entry systems in facilitating medication errors
.
JAMA
.
2005
;
293
(
10
):
1197
1203
20
Meeker
D
,
Knight
TK
,
Friedberg
MW
, et al
.
Nudging guideline-concordant antibiotic prescribing: a randomized clinical trial
.
JAMA Intern Med
.
2014
;
174
(
3
):
425
431
21
Schuler
CL
,
Courter
JD
,
Conneely
SE
, et al
.
Decreasing duration of antibiotic prescribing for uncomplicated skin and soft tissue infections
.
Pediatrics
.
2016
;
137
(
2
):
e20151223
22
Palms
DL
,
Hicks
LA
,
Bartoces
M
, et al
.
Comparison of antibiotic prescribing in retail clinics, urgent care centers, emergency departments, and traditional ambulatory care settings in the United States
.
JAMA Intern Med
.
2018
;
178
(
9
):
1267
1269

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.