Improving Antimicrobial-Prescribing in Emergency Departments

• Sponsor: Iowa City Veterans Affairs Medical Center (U.S. Fed)
• Overall Status: Completed
• CT.gov ID: NCT03349567
• Collaborator: Merck Sharp & Dohme LLC (Industry)
• Enrollment: 27
• Locations: 2
• Arms: 2
• Actual Duration (Months): 12.7
• Patients Per Site: 13.5
• Patients Per Site Per Month: 1.1

Study Details

Study Description

Brief Summary

Antimicrobial resistance is one of today's most urgent public health problems. One of the most important strategies to slow the spread of antimicrobial resistance is the promotion of judicious antimicrobial use. There are tremendous opportunities to reduce unnecessary antimicrobial-prescribing, particularly in Emergency Departments (EDs). In this study, the investigators will work collaboratively with ED providers in the Veterans Health Administration (VHA) to reduce unnecessary antimicrobial use. Academic-detailing and an audit-and-feedback intervention will be implemented, and the study will assess how overall antimicrobial-prescribing changes once these interventions are performed. ED providers will be shown how their antimicrobial-prescribing compares to their peers, thereby encouraging them to consider their professional reputation when making prescribing decisions. To assess the impact of this intervention, the study will monitor providers' antimicrobial-prescribing behavior through an automated metric, i.e. number of antimicrobial prescriptions per number of patient-visits. To assess changes in the appropriateness of antimicrobial-prescribing, the study team will also perform manual chart reviews and compare prescribing decisions to published guidelines.

Condition or Disease	Intervention/Treatment	Phase
Anti-Bacterial Agents; Respiratory Tract Infections	Behavioral: Audit-and-feedback	N/A

Detailed Description

Using a quasi-experimental design, the study team will work collaboratively with ED providers in the Veterans Health Administration (VHA) to improve antimicrobial-prescribing. Academic-detailing and an audit-and-feedback intervention will be implemented. Through audit-and-feedback, providers will be shown how their antimicrobial-prescribing compares to their peers, thereby encouraging them to consider their professional reputation when making prescribing decisions.

The study will use a pretest-posttest design (quasi-experimental design with a non-equivalent control group) to assess the effect of our pilot intervention at 2 participating EDs and 2 control EDs. For each ED, the pretest period will be the 12 months prior to the intervention. The intervention itself will last 12 months.

There will be an estimated 30 ED providers at the 2 study sites and 30 ED providers at the 2 control sites. Only antimicrobial prescriptions to patients discharged from the ED will be evaluated.

The primary outcome will be antimicrobial-prescribing volume adjusted for the total number of patient-visits. Secondary outcomes include the frequency of guideline-discordant antimicrobial-prescribing and adverse events.

At the conclusion of the pilot trial, a quasi-experimental interrupted time-series (ITS) analysis will be performed to assess the change in monthly antimicrobial usage for the 2 intervention sites combined and the 2 control sites combined. The time frame for this ITS analysis will be the one-year prior to the pilot trial's initiation through the trial's 1-year intervention period for a total of 24 months. In addition, a multivariable analysis will be performed to identify predictors of guideline-discordant therapy.

Study Design

• Study Type: Interventional
• Actual Enrollment: 27 participants
• Allocation: Non-Randomized
• Intervention Model: Parallel Assignment
• Masking: None (Open Label)
• Primary Purpose: Health Services Research
• Official Title: The Use of Audit-and-Feedback to Improve Antimicrobial-Prescribing in Emergency Departments: a Quasi-experimental Study
• Actual Study Start Date: Oct 9, 2018
• Actual Primary Completion Date: Sep 30, 2019
• Actual Study Completion Date: Oct 31, 2019

Arms and Interventions

Arm	Intervention/Treatment
Experimental: Audit-and-feedback The experimental arm will consist of Emergency Department providers who do receive the intervention.	Behavioral: Audit-and-feedback; We will monitor the antimicrobial-prescribing of providers in the experimental arm. Our study team will meet with providers in the experimental arm to provide guidance on optimal antimicrobial-prescribing. We will provide personalized feedback to providers in the experimental arm once every quarter.; Other Names: Academic detailing
No Intervention: Control The control arm will consist of providers who do not receive the intervention.

Outcome Measures

Primary Outcome Measures

1. Percentage of Patient-visits Prescribed an Antimicrobial at Their ED Visit [This outcome will be calculated every month during the baseline and intervention periods. An interrupted time-series analysis will be performed to evaluate how this monthly percentage changed after the intervention period began.]

   This outcomes is defined as the total number of antimicrobial-prescribing visits divided by the total number of patient-visits. This metric will be calculated on a monthly basis for each site (2 intervention, 2 control). It will be calculated during the baseline and intervention periods. Therefore, each site will contribute 24 data points.

Secondary Outcome Measures

1. Percentage of Patient-visits Prescribed a Late Antimicrobial Prescription More Than 24 Hours After But Within 30 Days of Their ED Visit [For both intervention and control sites, the outcome will be reported as a monthly median, which is calculated across the 12 baseline and the 12 intervention months.]

   Late antimicrobial prescription is defined as the prescription of an (additional) antimicrobial by any provider more than 24 hours after but within 30 days of the patient's ED visit.

2. Percentage of Patient-visits Who Were Hospitalized More Than 24 Hours After But Within 30 Days of Their ED Visit [For both intervention and control sites, the outcome will be reported as a monthly median, which is calculated across the 12 baseline and the 12 intervention months.]

   Hospitalization is defined as the need for hospitalization for any indication at a VA-facility more than 24 hours after but within 30 days of the patient's index visit to the ED.

3. Percentage of Patient-visits Who Died More Than 24 Hours After But Within 30 Days of Their ED Visit [For both intervention and control sites, the outcome will be reported as a monthly median, which is calculated across the 12 baseline and the 12 intervention months.]

   Death is defined as all-cause death more than 24 hours after and within 30-days of the patient's ED visit.

4. Percentage of Patient-visits Who Underwent Clostridium Difficile Testing More Than 24 Hours After But Within 30 Days of Their ED Visit [For both intervention and control sites, the outcome will be reported as a monthly median, which is calculated across the 12 baseline and the 12 intervention months.]

   Clostridium difficile testing is defined as a laboratory test for Clostridium difficile ordered more than 24 hours after but within 30-days of the patient's ED visit.

5. Percentage of Patient-visits Who Received Guideline Concordant Antimicrobial Management [For the baseline and intervention periods, this outcome will be reported as the percentage of patients who received guideline-concordant management among the sample of qualifying ED visits that were randomly selected for manual chart review.]

   This outcome was assessed using blinded manual chart reviews in a randomly selected sample of patients that met inclusion criteria, as outlined in our study protocol. For this reason, the denominators for this outcome will not match the patient counts for the other outcomes. If a patient received an antimicrobial when indicated or did not receive an antimicrobial when not indicated, the patient was deemed to have received guideline-concordant management. During the baseline period, 497 charts were randomly selected for review, and 480 ultimately underwent adjudication. During the intervention period, 576 charts were randomly selected for review, and 455 underwent adjudication.

6. Percentage of Patient-visits Prescribed an Antibiotic at Their ED Visit [This outcome will be calculated every month during the baseline and intervention periods. An interrupted time-series analysis will be performed to evaluate how this monthly percentage changed after the intervention period began.]

   This outcomes is defined as the total number of antibacterial-prescribing visits divided by the total number of patient-visits. This metric will be calculated on a monthly basis for each site (2 intervention, 2 control). It will be calculated during the baseline and intervention periods. Therefore, each site will contribute 24 data points.

Eligibility Criteria

Criteria

• Ages Eligible for Study: N/A and Older
• Sexes Eligible for Study: All
• Accepts Healthy Volunteers: Yes

Inclusion Criteria:

--An ED provider at one of the intervention or control sites.

Exclusion Criteria:

--An ED provider who sees less than 100 patients in the ED per year.

Contacts and Locations

Locations

	Site	City	State	Country	Postal Code
1	Richard Roudebush VAMC	Indianapolis	Indiana	United States	46202
2	Sioux Falls VA	Sioux Falls	South Dakota	United States	57105

Sponsors and Collaborators

• Iowa City Veterans Affairs Medical Center
• Merck Sharp & Dohme LLC

Investigators

None specified.

Study Documents (Full-Text)

• Study Protocol and Statistical Analysis Plan - Dec 24, 2020

More Information

Publications

None provided.

• Responsible Party: Daniel Livorsi, Physician-investigator, Iowa City Veterans Affairs Medical Center
• ClinicalTrials.gov Identifier: NCT03349567
• Other Study ID Numbers: 201708772
• First Posted: Nov 21, 2017
• Last Update Posted: Apr 28, 2021
• Last Verified: Apr 1, 2021

• Individual Participant Data (IPD) Sharing Statement: No
• Plan to Share IPD: No
• Studies a U.S. FDA-regulated Drug Product: No
• Studies a U.S. FDA-regulated Device Product: No
• Additional relevant MeSH terms: Respiratory Tract Infections; Emergencies

Study Results

Participant Flow

Recruitment Details

We performed a quasi-experimental study with an interrupted time-series design and a matched-pair non-equivalent control group. There were 2 intervention sites and 2 control sites; 27/31 providers were enrolled at intervention sites. We measured outcomes in the PATIENTS seen at these 4 EDs. In a difference-in-differences analysis, outcomes were compared across intervention and control sites in patients seen before the intervention period (i.e. baseline) to those during the intervention period.

Arm/Group Title	Audit-and-feedback	Control
Arm/Group Description	The experimental arm will consist of 2 Emergency Departments. We will attempt to enroll all ED providers at our 2 intervention sites, and for the analysis, site-level data will be used. Audit-and-feedback: We will monitor the antimicrobial-prescribing of providers in the experimental arm. Our study team will meet with providers in the experimental arm to provide guidance on optimal antimicrobial-prescribing. We will provide personalized feedback to providers in the experimental arm once every quarter.	The control arm will consist of 2 ED sites that do not receive the intervention. For the analysis, the control sites will be analyzed at the site-level.
Period Title: Overall Study
STARTED	2	2
COMPLETED	2	2
NOT COMPLETED	0	0

Baseline Characteristics

Arm/Group Title	Patients Seen in the ED at Audit-and-feedback Sites During the Intervention Period	Patients Seen in the ED at Control Sites During the Intervention Period	Patients Seen in the ED at Audit-and-feedback Sites During the Baseline Period	Patients Seen in the ED at Control Sites During the Baseline Period	Total
Arm/Group Description	These measures reflect patients seen at audit-and-feedback sites during the intervention period.	These baseline measures reflect patients seen at control sites during the intervention period.	These measures reflect patients seen at audit-and-feedback sites before the intervention period, i.e. during the baseline period.	These measures reflect patients seen at control sites before the intervention period, i.e. during the baseline period.	Total of all reporting groups
Overall Participants	23164	28835	28016	33077	113092
Age (years) [Median (Inter-Quartile Range) ]
Median (Inter-Quartile Range) [years]	60	60	60	60	60
Sex: Female, Male (Count of Participants)
Female	2532 10.9%	2627 9.1%	3014 10.8%	3159 9.6%	11332 10%
Male	20,632 NaN	26208 90.9%	25002 89.2%	29918 90.4%	101760 90%
Race (NIH/OMB) (Count of Participants)
American Indian or Alaska Native	204 0.9%	336 1.2%	243 0.9%	361 1.1%	1144 1%
Asian	39 0.2%	190 0.7%	70 0.2%	218 0.7%	517 0.5%
Native Hawaiian or Other Pacific Islander	55 0.2%	125 0.4%	94 0.3%	158 0.5%	432 0.4%
Black or African American	5106 22%	2720 9.4%	5990 21.4%	3215 9.7%	17031 15.1%
White	16343 70.6%	23832 82.6%	19893 71%	27323 82.6%	87391 77.3%
More than one race	321 1.4%	403 1.4%	389 1.4%	467 1.4%	1580 1.4%
Unknown or Not Reported	1096 4.7%	1229 4.3%	1337 4.8%	1335 4%	4997 4.4%
% of total visits prescribed an antimicrobial (monthly median) (Percentage of visits prescribed an abx) [Median (Inter-Quartile Range) ]
Median (Inter-Quartile Range) [Percentage of visits prescribed an abx]	19.1	21.6	21.2	20.9	20.7

Outcome Measures

1. Primary Outcome

Title	Percentage of Patient-visits Prescribed an Antimicrobial at Their ED Visit
Description	This outcomes is defined as the total number of antimicrobial-prescribing visits divided by the total number of patient-visits. This metric will be calculated on a monthly basis for each site (2 intervention, 2 control). It will be calculated during the baseline and intervention periods. Therefore, each site will contribute 24 data points.
Time Frame	This outcome will be calculated every month during the baseline and intervention periods. An interrupted time-series analysis will be performed to evaluate how this monthly percentage changed after the intervention period began.

Outcome Measure Data

Analysis Population Description

For the Outcome Measure Data Table shown below, patient-visits were stratified by period (i.e. baseline and intervention period), and outcomes were compared across these two periods between intervention and control sites. Please see our statistical analysis plan for further details. For the intervention and control sites, the "overall number of participants analyzed" in the Arms/Groups table equals the sum of the "number analyzed" for the baseline and intervention periods, as shown below.

Arm/Group Title	Patients Seen in the ED at Audit-and-feedback Sites	Patients Seen in the ED at Control Sites
Arm/Group Description	All patients seen in the ED at audit-and-feedback sites will be considered to have been exposed during the intervention.	All patients seen in the ED at audit-and-feedback sites will be considered to not have been exposed to the intervention.
Measure Participants	51180	61,912
Baseline	21.2	20.9
Intervention period	19.1	21.6

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Patients Seen in the ED at Audit-and-feedback Sites, Patients Seen in the ED at Control Sites
	Comments
	Type of Statistical Test	Superiority
	Comments
Statistical Test of Hypothesis	p-Value	0.35
	Comments
	Method	Mixed Models Analysis
	Comments	Segmented regression analysis was conducted using generalized linear models to estimate change in monthly antimicrobial prescription rates.
Method of Estimation	Estimation Parameter	Incident rate ratio
	Estimated Value	0.99
	Confidence Interval	(2-Sided) 95%; 0.98 to 1.01
	Parameter Dispersion	Type: Value:
	Estimation Comments

2. Secondary Outcome

Title	Percentage of Patient-visits Prescribed a Late Antimicrobial Prescription More Than 24 Hours After But Within 30 Days of Their ED Visit
Description	Late antimicrobial prescription is defined as the prescription of an (additional) antimicrobial by any provider more than 24 hours after but within 30 days of the patient's ED visit.
Time Frame	For both intervention and control sites, the outcome will be reported as a monthly median, which is calculated across the 12 baseline and the 12 intervention months.

Outcome Measure Data

Analysis Population Description

For the Outcome Measure Data Table shown below, patient-visits were stratified by period (i.e. baseline and intervention period), and outcomes were compared across these two periods between intervention and control sites. Please see our statistical analysis plan for further details. For the intervention and control sites, the "overall number of participants analyzed" in the Arms/Groups table equals the sum of the "number analyzed" for the baseline and intervention periods, as shown below.

Arm/Group Title	Patients Seen in the ED at Audit-and-feedback Sites	Patients Seen in the ED at Control Sites
Arm/Group Description	All patients seen in the ED at audit-and-feedback sites will be considered to have been exposed during the intervention.	All patients seen in the ED at audit-and-feedback sites will be considered to not have been exposed to the intervention.
Measure Participants	51180	61912
Baseline	31.5	29.5
Intervention	31.1	28.5

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Patients Seen in the ED at Audit-and-feedback Sites, Patients Seen in the ED at Control Sites
	Comments
	Type of Statistical Test	Non-Inferiority
	Comments	A Poisson regression model with log link was used to estimate the rate of safety outcomes. Models were adjusted for time as a continuous covariate, an indicator for month of implementation of intervention (Oct 2018), and included random effects to account for repeated measurements. An interaction variable for the study period (baseline or intervention) and site (intervention or control) was included in the model to estimate the Incident Rate Ratio (IRR) and 95% CIs.
Statistical Test of Hypothesis	p-Value
	Comments
	Method
	Comments
Method of Estimation	Estimation Parameter	Incident rate ratio
	Estimated Value	1.12
	Confidence Interval	(2-Sided) 95%; 0.93 to 1.35
	Parameter Dispersion	Type: Value:
	Estimation Comments

3. Secondary Outcome

Title	Percentage of Patient-visits Who Were Hospitalized More Than 24 Hours After But Within 30 Days of Their ED Visit
Description	Hospitalization is defined as the need for hospitalization for any indication at a VA-facility more than 24 hours after but within 30 days of the patient's index visit to the ED.
Time Frame	For both intervention and control sites, the outcome will be reported as a monthly median, which is calculated across the 12 baseline and the 12 intervention months.

Outcome Measure Data

Analysis Population Description

For the Outcome Measure Data Table shown below, patient-visits were stratified by period (i.e. baseline and intervention period), and outcomes were compared across these two periods between intervention and control sites. Please see our statistical analysis plan for further details. For the intervention and control sites, the "overall number of participants analyzed" in the Arms/Groups table equals the sum of the "number analyzed" for the baseline and intervention periods, as shown below.

Arm/Group Title	Patients Seen in the ED at Audit-and-feedback Sites	Patients Seen in the ED at Control Sites
Arm/Group Description	All patients seen in the ED at audit-and-feedback sites will be considered to have been exposed during the intervention.	All patients seen in the ED at audit-and-feedback sites will be considered to not have been exposed to the intervention.
Measure Participants	51180	61912
Baseline	5.9	5.6
Intervention period	5.9	5.8

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Patients Seen in the ED at Audit-and-feedback Sites, Patients Seen in the ED at Control Sites
	Comments
	Type of Statistical Test	Non-Inferiority
	Comments	A Poisson regression model with log link was used to estimate the rate of safety outcomes. Models were adjusted for time as a continuous covariate, an indicator for month of implementation of intervention (Oct 2018), and included random effects to account for repeated measurements. An interaction variable for the study period (baseline or intervention) and site (intervention or control) was included in the model to estimate the Incident Rate Ratio (IRR) and 95% CIs.
Statistical Test of Hypothesis	p-Value
	Comments
	Method
	Comments
Method of Estimation	Estimation Parameter	Incident rate ratio
	Estimated Value	1.01
	Confidence Interval	(2-Sided) 95%; 0.81 to 1.27
	Parameter Dispersion	Type: Value:
	Estimation Comments

4. Secondary Outcome

Title	Percentage of Patient-visits Who Died More Than 24 Hours After But Within 30 Days of Their ED Visit
Description	Death is defined as all-cause death more than 24 hours after and within 30-days of the patient's ED visit.
Time Frame	For both intervention and control sites, the outcome will be reported as a monthly median, which is calculated across the 12 baseline and the 12 intervention months.

Outcome Measure Data

Analysis Population Description

For the Outcome Measure Data Table shown below, patient-visits were stratified by period (i.e. baseline and intervention period), and outcomes were compared across these two periods between intervention and control sites. Please see our statistical analysis plan for further details. For the intervention and control sites, the "overall number of participants analyzed" in the Arms/Groups table equals the sum of the "number analyzed" for the baseline and intervention periods, as shown below.

Arm/Group Title	Patients Seen in the ED at Audit-and-feedback Sites	Patients Seen in the ED at Control Sites
Arm/Group Description	All patients seen in the ED at audit-and-feedback sites will be considered to have been exposed during the intervention.	All patients seen in the ED at audit-and-feedback sites will be considered to not have been exposed to the intervention.
Measure Participants	51180	61912
Baseline	0.56	0.61
Intervention period	0.60	0.59

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Patients Seen in the ED at Audit-and-feedback Sites, Patients Seen in the ED at Control Sites
	Comments
	Type of Statistical Test	Non-Inferiority
	Comments	A Poisson regression model with log link was used to estimate the rate of safety outcomes. Models were adjusted for time as a continuous covariate, an indicator for month of implementation of intervention (Oct 2018), and included random effects to account for repeated measurements. An interaction variable for the study period (baseline or intervention) and site (intervention or control) was included in the model to estimate the Incident Rate Ratio (IRR) and 95% CIs.
Statistical Test of Hypothesis	p-Value
	Comments
	Method
	Comments
Method of Estimation	Estimation Parameter	Incident rate ratio
	Estimated Value	0.88
	Confidence Interval	(2-Sided) 95%; 0.58 to 1.34
	Parameter Dispersion	Type: Value:
	Estimation Comments

5. Secondary Outcome

Title	Percentage of Patient-visits Who Underwent Clostridium Difficile Testing More Than 24 Hours After But Within 30 Days of Their ED Visit
Description	Clostridium difficile testing is defined as a laboratory test for Clostridium difficile ordered more than 24 hours after but within 30-days of the patient's ED visit.
Time Frame	For both intervention and control sites, the outcome will be reported as a monthly median, which is calculated across the 12 baseline and the 12 intervention months.

Outcome Measure Data

Analysis Population Description

For the Outcome Measure Data Table shown below, patient-visits were stratified by period (i.e. baseline and intervention period), and outcomes were compared across these two periods between intervention and control sites. Please see our statistical analysis plan for further details. For the intervention and control sites, the "overall number of participants analyzed" in the Arms/Groups table equals the sum of the "number analyzed" for the baseline and intervention periods, as shown below.

Arm/Group Title	Patients Seen in the ED at Audit-and-feedback Sites	Patients Seen in the ED at Control Sites
Arm/Group Description	All patients seen in the ED at audit-and-feedback sites will be considered to have been exposed during the intervention.	All patients seen in the ED at audit-and-feedback sites will be considered to not have been exposed to the intervention.
Measure Participants	51180	61912
Baseline	0.60	0.86
Intervention period	0.69	0.89

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Patients Seen in the ED at Audit-and-feedback Sites, Patients Seen in the ED at Control Sites
	Comments
	Type of Statistical Test	Non-Inferiority
	Comments	A Poisson regression model with log link was used to estimate the rate of safety outcomes. Models were adjusted for time as a continuous covariate, an indicator for month of implementation of intervention (Oct 2018), and included random effects to account for repeated measurements. An interaction variable for the study period (baseline or intervention) and site (intervention or control) was included in the model to estimate the Incident Rate Ratio (IRR) and 95% CIs.
Statistical Test of Hypothesis	p-Value
	Comments
	Method
	Comments
Method of Estimation	Estimation Parameter	Incident rate ratio
	Estimated Value	0.83
	Confidence Interval	(2-Sided) 95%; 0.53 to 1.29
	Parameter Dispersion	Type: Value:
	Estimation Comments

6. Secondary Outcome

Title	Percentage of Patient-visits Who Received Guideline Concordant Antimicrobial Management
Description	This outcome was assessed using blinded manual chart reviews in a randomly selected sample of patients that met inclusion criteria, as outlined in our study protocol. For this reason, the denominators for this outcome will not match the patient counts for the other outcomes. If a patient received an antimicrobial when indicated or did not receive an antimicrobial when not indicated, the patient was deemed to have received guideline-concordant management. During the baseline period, 497 charts were randomly selected for review, and 480 ultimately underwent adjudication. During the intervention period, 576 charts were randomly selected for review, and 455 underwent adjudication.
Time Frame	For the baseline and intervention periods, this outcome will be reported as the percentage of patients who received guideline-concordant management among the sample of qualifying ED visits that were randomly selected for manual chart review.

Outcome Measure Data

Analysis Population Description

For the Outcome Measure Data Table shown below, patient-visits were stratified by period (i.e. baseline and intervention period), and outcomes were compared across these two periods at intervention and control sites. Please see our statistical analysis plan for further details. For the intervention and control sites, the "overall number of participants analyzed" in the Arms/Groups table equals the sum of the "number analyzed" for the baseline and intervention periods, as shown below.

Arm/Group Title	Patients Seen in the ED at Audit-and-feedback Sites	Patients Seen in the ED at Control Sites
Arm/Group Description	All patients seen in the ED at audit-and-feedback sites will be considered to have been exposed during the intervention.	All patients seen in the ED at audit-and-feedback sites will be considered to not have been exposed to the intervention.
Measure Participants	470	465
Baseline	125 0.5%	123 0.4%
Intervention	166 0.7%	131 0.5%

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Patients Seen in the ED at Audit-and-feedback Sites
	Comments
	Type of Statistical Test	Superiority
	Comments
Statistical Test of Hypothesis	p-Value	0.001
	Comments
	Method	Chi-squared
	Comments

Statistical Analysis 2

Statistical Analysis Overview	Comparison Group Selection	Patients Seen in the ED at Control Sites
	Comments
	Type of Statistical Test	Superiority
	Comments
Statistical Test of Hypothesis	p-Value	0.13
	Comments
	Method	Chi-squared
	Comments

7. Secondary Outcome

Title	Percentage of Patient-visits Prescribed an Antibiotic at Their ED Visit
Description	This outcomes is defined as the total number of antibacterial-prescribing visits divided by the total number of patient-visits. This metric will be calculated on a monthly basis for each site (2 intervention, 2 control). It will be calculated during the baseline and intervention periods. Therefore, each site will contribute 24 data points.
Time Frame	This outcome will be calculated every month during the baseline and intervention periods. An interrupted time-series analysis will be performed to evaluate how this monthly percentage changed after the intervention period began.

Outcome Measure Data

Analysis Population Description

For the Outcome Measure Data Table shown below, patient-visits were stratified by period (i.e. baseline and intervention period), and outcomes were compared across these two periods between intervention and control sites. Please see our statistical analysis plan for further details. For the intervention and control sites, the "overall number of participants analyzed" in the Arms/Groups table equals the sum of the "number analyzed" for the baseline and intervention periods, as shown below.

Arm/Group Title	Patients Seen in the ED at Audit-and-feedback Sites	Patients Seen in the ED at Control Sites
Arm/Group Description	All patients seen in the ED at audit-and-feedback sites will be considered to have been exposed during the intervention.	All patients seen in the ED at audit-and-feedback sites will be considered to not have been exposed to the intervention.
Measure Participants	51180	61912
Baseline	20.8	20.6
Intervention	18.2	20.6

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Patients Seen in the ED at Audit-and-feedback Sites, Patients Seen in the ED at Control Sites
	Comments
	Type of Statistical Test	Superiority
	Comments
Statistical Test of Hypothesis	p-Value	0.46
	Comments
	Method	Regression, Logistic
	Comments	Segmented regression analysis was conducted using generalized linear models to estimate change in monthly antibiotic prescription rates.

Adverse Events

Time Frame	Outcomes were tracked for 30 days after each ED patient visit.
Adverse Event Reporting Description	Other adverse events consisted of Clostridioides difficile infections
Arm/Group Title	Patients Seen at Audit-and-feedback ED Sites During Intervention Period		Patients Seen at Control ED Sites During the Intervention Period		Patients Seen at Audit-and-feedback ED Sites During the Baseline Period		Patients Seen at Control EDs During the Baseline Period
Arm/Group Description	All patients seen in the EDs undergoing the audit-and-feedback intervention will be considered to have been exposed to the intervention. For this adverse event reporting section, we are reporting ED patient-visits during the intervention period at intervention sites. Because this study was not randomized, a more accurate assessment of adverse events would require a difference-in-differences analysis across the baseline and intervention periods while controlling for baseline differences between intervention and control sites. Such an analysis is described in our statistical analysis plan and will be reported in our published paper.		All patients seen in the EDs designated as control sites will be considered to have been part of the control group. For this adverse event reporting section, we are reporting ED patient-visits during the intervention period at control sites. Because this study was not randomized, a more accurate assessment of adverse events would require a difference-in-differences analysis across the baseline and intervention periods while controlling for baseline differences between intervention and control sites. Such an analysis is described in our statistical analysis plan and will be reported in our published paper.		For this adverse event reporting section, we are only reporting ED patient-visits at intervention sites during the baseline period.		For this adverse event reporting section, we are reporting ED patient-visits at control sites during the baseline period.
All Cause Mortality
	Patients Seen at Audit-and-feedback ED Sites During Intervention Period		Patients Seen at Control ED Sites During the Intervention Period		Patients Seen at Audit-and-feedback ED Sites During the Baseline Period		Patients Seen at Control EDs During the Baseline Period
	Affected / at Risk (%)	# Events	Affected / at Risk (%)	# Events	Affected / at Risk (%)	# Events	Affected / at Risk (%)	# Events
Total	141/23164 (0.6%)		176/28835 (0.6%)		146/28016 (0.5%)		197/33077 (0.6%)
Serious Adverse Events
	Patients Seen at Audit-and-feedback ED Sites During Intervention Period		Patients Seen at Control ED Sites During the Intervention Period		Patients Seen at Audit-and-feedback ED Sites During the Baseline Period		Patients Seen at Control EDs During the Baseline Period
	Affected / at Risk (%)	# Events	Affected / at Risk (%)	# Events	Affected / at Risk (%)	# Events	Affected / at Risk (%)	# Events
Total	1593/23164 (6.9%)		1863/28835 (6.5%)		1810/28016 (6.5%)		2031/33077 (6.1%)
General disorders
Hospitalization >24 hours after but within 30 days of ED visit	1452/23164 (6.3%)	1452	1687/28835 (5.9%)	1687	1664/28016 (5.9%)	1664	1834/33077 (5.5%)	1834
Mortality > 24 hours after ED visit and within 30 days of visit	141/23164 (0.6%)	141	176/28835 (0.6%)	176	146/28016 (0.5%)	146	197/33077 (0.6%)	197
Other (Not Including Serious) Adverse Events
	Patients Seen at Audit-and-feedback ED Sites During Intervention Period		Patients Seen at Control ED Sites During the Intervention Period		Patients Seen at Audit-and-feedback ED Sites During the Baseline Period		Patients Seen at Control EDs During the Baseline Period
	Affected / at Risk (%)	# Events	Affected / at Risk (%)	# Events	Affected / at Risk (%)	# Events	Affected / at Risk (%)	# Events
Total	11/23164 (0%)		27/28835 (0.1%)		193/28016 (0.7%)		321/33077 (1%)
Gastrointestinal disorders
Clostridioides difficile infection	11/23164 (0%)	11	27/28835 (0.1%)	27	193/28016 (0.7%)	28016	321/33077 (1%)	33077

Limitations/Caveats

Because the study was not randomized, we cannot exclude temporal confounding or selection bias. By conducting a time-series analysis, we tried to account for confounders that may change over time, and by including a matched-control group, we tried to minimize the possibility of regression to the mean. Second, we cannot exclude the possibility that sites had any other processes in place that could have influenced our outcomes.

More Information

Certain Agreements

Principal Investigators are NOT employed by the organization sponsoring the study.

There is NOT an agreement between Principal Investigators and the Sponsor (or its agents) that restricts the PI's rights to discuss or publish trial results after the trial is completed.