Audit and Feedback (A&F), a strategy aimed at promoting modified practice through performance feedback, is a method to change provider behaviour and reduce unnecessary medical services. This study aims to assess the use of A&F to change:

1. antibiotic prescribing for viral infections
2. antipsychotic prescribing to patients with dementia
3. routine measure of vitamin D in low risk adults
4. annual screening blood tests (without direct indication by the risk profile of the patient)

The investigators conducted a clustered randomized trial of primary care providers in Manitoba, Canada participating in the Manitoba Primary Care Research Network (MaPCReN) to assess the impact of A&F interventions to change prescribing and screening rates.

In January 2016, MaPCReN included 239 providers in 46 practices across Manitoba, representing ~20% of Manitoba primary care providers. MaPCReN recruits consenting primary care providers (family physicians, nurse practitioners and community pediatricians) that receive semi-annual feedback reports for practice reflection and quality improvement. Feedback reports provide practice-level details including disease prevalence and clinical characteristics of patients, reported in comparison to other providers at the same practice, health region and provincially

Investigators implemented an A&F cycle reporting on practice behaviours related to four Choosing Wisely Canada (CWC) recommendations.

A) "Don't use antibiotics for upper respiratory infections that are likely viral in origin, such as influenza-like illness, or self-limiting, such as sinus infections of less than seven days of duration." • Bacterial infections of the respiratory tract are rare and usually only secondary complications from initial viral infections. Antibiotics are rarely indicated for upper respiratory illness and patients may suffer adverse effects from antibiotics.

B) Don't routinely measure Vitamin D in low risk adults.

• Routine supplementation of vitamin D is recommended for people living in Northern climates regardless of measured laboratory levels. There is no indication for ordering vitamin D unless a patient is suffering severe renal or metabolic disease.

C) Don't do annual screening blood tests unless directly indicated by the risk profile of the patient.

• There is minimal value in ordering routine blood tests in asymptomatic patients and in fact some "routine" screening tests such as prostate-specific antigen (PSA) is in fact recommended against due to the likelihood of producing false positive results. This may lead to additional and unnecessary testing.

D) Don't prescribe antipsychotics to treat behavioural and psychological symptoms of dementia.

• There are some instances where patients with behavioural symptoms of dementia require antipsychotic medicines although these instances are rare. These medicines should be used judiciously given for this indication because they provide limited benefit and may cause serious side effects including premature death

MaPCReN personnel randomly assigned numbers to clinics for randomization. Number of physicians practicing at each clinic was controlled during randomization.

The intervention "Group 1" received the standard feedback report with a one-page summary of the CWC recommendations of interest. Group 1 did not receive any data specific to their prescribing. Intervention "Group 2" received the standard feedback report, CWC recommendation summary and practice-specific data related to their prescribing rates for the CWC recommendations of interest, compared to rates for other providers at their clinic, in their health region and in the province. The control group received the standard feedback report with no information related to CWC.

The A&F intervention took place between January 2016 and January 2017. Retrospective Electronic Medical Record (EMR) data from the MaPCReN repository was assessed between January 2014-December 2019 to assess changes in prescribing rates during and immediately following the A&F intervention (2016/17), as well as assess continued change in practice after the A&F (2018/19) compared to prescribing rates prior to the A&F (2014/15). There were seven clinics not included in the 2018/19 analysis due to a change in EMR vendor that may have compromised data quality.

Inclusion Criteria:

• All primary care providers participating in the Manitoba Primary Care Research Network (MaPCReN), the Manitoba practice-based network of the Canadian Primary Care Sentinel Surveillance Network.

Exclusion Criteria:

• Primary care providers not participating in MaPCReN

Clinics were randomized to one of three groups. Electronic medical records were extracted for each group. All extracted EMR data is de-identified included the patient, provider and site details. Using a unique site identification number, sites were linked to their randomization group for analysis.