CMO Letter to Reduce Inappropriate Antibiotic Prescribing Winter 2019/2020

A Letter From the CMO in England to Reduce Antibiotic Prescribing in General Practice: A Randomized Controlled Trial Comparing Monitoring and Feedback (With and Without Graphs), and Social Norms Feedback (Without and Without a Specific Case Study of Patient Harm)

This trial aims to reduce inappropriate prescription of antibiotics and broad spectrum antibiotics by general practitioners (GPs) in England. Unnecessary prescriptions are defined as those that do not improve patient health outcomes. The intervention is to send GPs a letter from the Chief Medical Officer (CMO) that gives feedback on their practice's prescribing levels.

There will be three intervention samples:

1. practices whose prescribing in the past year was under the new target of 0.965 items per STAR-PU but who would exceed the target if they had a 5% increase in prescribing; trial compares prescribing of practices whose GPs receive a letter informing them that their practice's prescribing is just under the new target to that of practices that are not sent a letter
2. Practices whose prescribing in the past year was above the new target but who not in the top 20% of prescribers; trial compares prescribing of practices whose GPs receive a letter informing them that their practice's prescribing exceeds the new target to practices who get a letter that includes a graph showing their prescribing relative to the target and to practices that are not sent a letter
3. Practices that are currently in the top 20% of prescribers; trial compares effect on prescribing of a feedback letter with a social norms message (current standard practice for this group) to a letter informing GPs that their practice's prescribing exceeds the new target and to a letter with a social norms message, that includes a specific example of a case of patient harm caused by antimicrobial resistance.

Study Overview

• Status: Completed
• Conditions: Prescribing, Off-Label
• Intervention / Treatment: Behavioral: Letter

Detailed Description

The study will involve three trials, each conducted as non-blinded randomised controlled trial, with GP practices as the unit of randomisation.

Trial 1 Targeting practices whose prescribing in the past year was under the new target but who would exceed the target if they had a 5% increase in prescribing

• Control: No letter
• Intervention: Letter informing them that their practice's prescribing is just under the new target (Letter A) Trial hypothesis: Sending a letter to GPs whose practices are just under the new prescribing target will reduce antibiotic prescribing

Trial 2 Targeting practices whose prescribing in the past year was above the new target but who not in the top 20% of prescribers

• Control: No letter
• Intervention 1: Letter informing them that their practice's prescribing exceeds the new target (Letter B1)
• Intervention 2: Letter informing them that their practice's prescribing exceeds the new target with a graph representing prescribing relative to the target (Letter B2) Hypotheses: (i) Sending a letter to GPs whose practices missed the new prescribing target will reduce their prescribing; (ii) A letter with a graph will be more effective than a letter without a graph.

Trial 3 Targeting practices that are currently in the top 20% of prescribers

• Control: Current standard practice, a social norms message, that their practice is in the top 20% of prescribers (Letter C1)
• Intervention 1: Letter informing them that their practice's prescribing exceeds the new target (Letter C2)
• Intervention 2: Social norms message, that they are in the top 20%, with a specific example of a case of patient harm caused by antimicrobial resistance (Letter C3)

Hypotheses: (i) A letter with a social norms message and a specific example of a case where a patient came to harm will be more effective than a feedback letter without a specific example; (ii) A letter telling GPs that they missed the prescribing target will be no less effective than a letter with social norms feedback

For each letter, there will be two versions, one for practices whose prescribing has increased by > 5% in the previous year, informing them of that their prescribing has increased since the previous year, and one for practices whose prescribing has not been increasing.

The letters will signpost GPs to resources to help address patient demand for inappropriate antibiotic prescribing, recognising that many GPs feel that patients expect antibiotics and that GPs may find it difficult to have the necessary patient conversations, especially within a short consultation. As with previous letters, these letters will advise GPs of actions that they can take to reduce inappropriate prescribing, supporting them to have conversations with patients, and there will be TARGET leaflets enclosed.

Power calculation All trials are powered to detect a 2% reduction in prescribing at a significance level of 0.05 with a power of 80%.

Statistical analysis plan In order to test our hypotheses, the investigators will use a fixed effects panel regression model, with time trends accounting for seasonal effects, to estimate the effect of treatment status on prescribing. The investigators will also run ANCOVAs for each month separately and one covering the whole six months of the trial. Analysis will control for baseline prescribing rates and for whether practices got the version of the letter saying that their prescribing has been increasing.

• Study Type: Interventional
• Enrollment (Actual): 2963
• Phase: Not Applicable

Contacts and Locations

Study Locations

• United Kingdom
  • London, United Kingdom, SE1
    • Public Health England

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Child; Adult; Older Adult
• Accepts Healthy Volunteers: No
• Genders Eligible for Study: All

Description

Inclusion Criteria:

• GP practices that prescribed more than 0.919 Antibacterial Items/STAR- PU (5% under the target of 0.965) for the twelve months April 2018 - March 2019

Exclusion Criteria:

• Practices in the 99th percentile of prescribers

Study Plan

How is the study designed?

Design Details

• Primary Purpose: Other
• Allocation: Randomized
• Interventional Model: Parallel Assignment
• Masking: Single

Number of Arms

8

Arms and Interventions

Participant Group / Arm	Intervention / Treatment
No Intervention: Just under target control; Practices whose prescribing in the past year was under the new target but who would exceed the target if they had a 5% increase in prescribing; no letter sent.
Experimental: Just under target letter; Practices whose prescribing in the past year was under the new target but who would exceed the target if they had a 5% increase in prescribing: receive a letter informing of this. Randomization is stratified according to whether their prescribing had increased by > 5% compared to the previous year; those whose prescribing had increased had it mentioned in the letter	Behavioral: Letter; Letters sent to GPs in relevant practices (prescribing data is by practice, so the practice is the unit of randomization)
No Intervention: Over target control; Practices whose prescribing in the past year was above the new target but who were not in the top 20% of prescribers; no letter sent; Intervention 1: Letter informing them that their practice's prescribing exceeds the new target (Letter B1); Intervention 2: Letter informing them that their practice's prescribing exceeds the new target with a graph representing prescribing relative to the target (Letter B2)
Experimental: Over target letter; Practices whose prescribing in the past year was above the new target but who were not in the top 20% of prescribers; receive a letter informing them that their practice's prescribing exceeds the new target (Letter B1) Randomization is stratified according to whether their prescribing had increased by > 5% compared to the previous year; those whose prescribing had increased had it mentioned in the letter	Behavioral: Letter; Letters sent to GPs in relevant practices (prescribing data is by practice, so the practice is the unit of randomization)
Experimental: Over target letter with bar chart; Practices whose prescribing in the past year was above the new target but who were not in the top 20% of prescribers; receive a letter informing them that their practice's prescribing exceeds the new target, including a bar chart showing their prescribing compared to the target (Letter B1) Randomization is stratified according to whether their prescribing had increased by > 5% compared to the previous year; those whose prescribing had increased had it mentioned in the letter	Behavioral: Letter; Letters sent to GPs in relevant practices (prescribing data is by practice, so the practice is the unit of randomization)
Active Comparator: Top 20% feedback letter control; Targeting practices that are currently in the top 20% of prescribers; letters informing them of the percentile they are on--standard practice--(Letter C1) Randomization is stratified according to whether their prescribing had increased by > 5% compared to the previous year; those whose prescribing had increased had it mentioned in the letter	Behavioral: Letter; Letters sent to GPs in relevant practices (prescribing data is by practice, so the practice is the unit of randomization)
Experimental: Top 20% above target letter; Targeting practices that are currently in the top 20% of prescribers; letters informing them that their prescribing exceeds the new target (Letter C2) Randomization is stratified according to whether their prescribing had increased by > 5% compared to the previous year; those whose prescribing had increased had it mentioned in the letter.	Behavioral: Letter; Letters sent to GPs in relevant practices (prescribing data is by practice, so the practice is the unit of randomization)
Experimental: Top 20% feedback letter with specific example of patient harm; Targeting practices that are currently in the top 20% of prescribers • Control: Current standard practice, a social norms message, that their practice is in the top 20% of prescribers (Letter C1) Targeting practices that are currently in the top 20% of prescribers; letters informing them of the percentile they are on with a specific example of a case of patient harm caused by antimicrobial resistance (Letter C3) Randomization is stratified according to whether their prescribing had increased by > 5% compared to the previous year; those whose prescribing had increased had it mentioned in the letter.	Behavioral: Letter; Letters sent to GPs in relevant practices (prescribing data is by practice, so the practice is the unit of randomization)

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Total antibiotic prescribing in September	antibiotic prescribing weighted by Specific Therapeutic group Age-sex Related Prescribing Unit (STAR-PU)	1 month
Total antibiotic prescribing in October	antibiotic prescribing weighted by Specific Therapeutic group Age-sex Related Prescribing Unit (STAR-PU)	2 months
Total antibiotic prescribing in November	antibiotic prescribing weighted by Specific Therapeutic group Age-sex Related Prescribing Unit (STAR-PU)	3 months
Total antibiotic prescribing in December	antibiotic prescribing weighted by Specific Therapeutic group Age-sex Related Prescribing Unit (STAR-PU)	4 months
Total antibiotic prescribing in January	antibiotic prescribing weighted by Specific Therapeutic group Age-sex Related Prescribing Unit (STAR-PU)	5 months
Total antibiotic prescribing in February	antibiotic prescribing weighted by Specific Therapeutic group Age-sex Related Prescribing Unit (STAR-PU)	6 months
Total antibiotic prescribing in from September-February	antibiotic prescribing weighted by Specific Therapeutic group Age-sex Related Prescribing Unit (STAR-PU)	6 months
Proportion of practices in each group whose prescribing was under the target	Whether antibiotic prescribing weighted by Specific Therapeutic group Age-sex Related Prescribing Unit (STAR-PU) for April 2019-March 2020 is under the NHS target of 0.965 items per STAR-PU	8 months

Collaborators and Investigators

• Sponsor: Public Health England

Study record dates

Study Major Dates

• Study Start (Actual): August 31, 2019
• Primary Completion (Actual): August 30, 2020
• Study Completion (Actual): August 30, 2020

Study Registration Dates

• First Submitted: August 7, 2019
• First Submitted That Met QC Criteria: August 7, 2019
• First Posted (Actual): August 9, 2019

Study Record Updates

• Last Update Posted (Actual): September 29, 2020
• Last Update Submitted That Met QC Criteria: September 28, 2020
• Last Verified: September 1, 2020

More Information

Terms related to this study

• Keywords: antimicrobial resistance; social norms feedback; behavioural science; feedback and monitoring
• Other Study ID Numbers: CMO Letter 2019/20

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: NO
• IPD Plan Description: Note that the trial will use publicly available prescribing data, so any researcher should be able to access it.

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No