Feedback to Improve Rational Strategies of Antibiotic Initiation and Duration in Long Term Care

Overarching Goals

The overarching goals of FIRST AID - LTC are two-fold:

• Improve rational antibiotic prescribing by physicians to minimize harms among LTC residents.
• Advance the science of audit-and-feedback to improve physician prescribing practices.

Specific Aims

To improve rational antibiotic prescribing in LTC:

• by decreasing unnecessary initiation of antibiotic treatments among Ontario LTC residents, as well as the variability in initiation rates across LTC prescribers.
• by decreasing unnecessary prolonged duration of antibiotic treatments among Ontario LTC residents, as well as the variability in prolonged duration treatment use across LTC prescribers.

To advance audit-and-feedback implementation science:

• by evaluating whether a dynamic audit-and-feedback report highlighting antibiotic prescribing can lead to greater reductions in antibiotic use, than a static paginated report

Anticipated Contributions to Health-Related Knowledge

Although the literature is inundated with trials examining the impact of audit-and-feedback compared to usual care, there is a need for studies to improve audit-and-feedback delivery. FIRST AID-LTC will test optimal delivery and peer comparison techniques for audit-and-feedback. The knowledge learned can be extrapolated to antibiotic interventions in LTC in other provinces across Canada, as well more broadly to inappropriate medication prescribing practices in LTC.

Anticipated Contributions to Health Care, Health Systems and Health Outcomes

FIRST AID-LTC will lead to immediate reductions in excess antibiotic use in Ontario LTC facilities, which in turn should result in substantial reductions in direct drug costs, as well as downstream complications of allergy, organ toxicity, C. difficile infections and antimicrobial resistance. With easy transferability to other Canadian provinces, the improvements in cost-savings and patient outcomes could be massive in scope.

To Identify an LTC Resident

Inclusion criteria

An individual having a minimum of 2 records on separate days within the quarter meeting any combination of the following criteria:

• a record for a non-emergency long-term care inpatient services OR
• an Ontario Drug Benefits record administered in long-term care

Index date = The analysis will be anchored on the most recent of either of the records above within a given quarter or their date of death (whichever date is earliest)

Exclusion criteria

• Non-Ontario resident at index date
• Invalid age (age<19 or age>115) at index date
• Missing or invalid sex or date of birth at index date
• Death date is >7 days before index date
• If the individual does not live in a nursing home or home for the aged
• Cannot be linked to a Most Responsible Physician (MRP) (see methodology below)

To Identify the Most Responsible Physician (MRP) Using Virtual Rostering

For each patient in the above resident cohort, the study team will retrieve all records from health care providers in the 6 month period preceding the index date (180 days), keeping only records from physicians who have a specialty of 1) general practice, 2) community medicine or 3) geriatrics.

Steps for MRP assignment:

Step 1) The study team will first select physicians with the highest count of records for the monthly management of a nursing home or home for the aged. This is completed for as many residents as possible.

Step 2) If there were no monthly management fee records as described above then the physician with highest count of non-emergency long-term care inpatient services records for each patient will be selected. This step is only applied to residents who could not be matched to a physician by Step 1. **Physician must have seen the patient one or more times in 90 days prior to and including index date to be considered MRP. This criteria is applied to ensure the physician has seen the resident within the reporting quarter.

Step 3) Some patients will virtually roster to physicians in Enrollment groups, some will virtually roster to physicians that are not in a group. For these, the study team will recode enrollment program type to 'NOR' (not otherwise rostered) - these are likely fee for service physicians.