MIDAS Cluster Randomized Controlled Trial of Implementation Strategies to Optimize Use of Medications in VA Clinical Settings

Percentage of PIMs across AD vs. LEAP+AD facilities were modelled as the difference between post-period and pre-period, using the average from the 1-6-month pre-baseline period as "pre" and the average 13-18-month post-baseline as "post." Data was collected monthly at patient level and collapsed by clinic-month for patients who are at risk of potentially inappropriate medication use. Clinic-month outcome was computed as: 1) VIONE; proportion of patients who possessed one or more medications from the Beers' list of patients 65 or older, actively following with the clinic, and not in hospice/palliative care; 2) DOACs; proportion of patients with flags for potentially inappropriate use on a DOAC safety dashboard of those using DOACs; 3) CBTI; proportion of patients with a new prescription for a sleep medication for insomnia who have not had CBTI of those who are actively followed by the clinic and not in hospice/palliative care. The outcome was analyzed by pooling across all three-EBPs.

Medications include proton pump inhibitors (PPIs), aspirin, central nervous system (CNS) active medications (muscle relaxants, anti-psychotics, Z-drugs, and benzodiazepines), or anticholinergic drugs. This is the primary outcome for the VIONE trial when analyzed as a stand-alone trial and the VIONE sub-analysis of the overall MIDAS study primary outcome. Percentage of patients with PIMs across AD vs. LEAP+AD facilities were modelled longitudinally using Generalized Estimating Equations (GEE), with a three way-interaction of arm, month of follow-up, and pre-, post-period. Data will was be collected monthly at patient level and collapsed by clinic-month for patients who are at risk of potentially inappropriate medication use. Estimates computed using least squares (LS) means from the GEE model.

Cost of all drugs without regard to appropriateness. Average monthly cost per patient across LEAP vs. LEAP + AD facilities were modelled longitudinally with a Generalized Linear Model with a Gamma Link and a three-way interaction of arm, month of follow-up and pre-, post-period. Estimates computed using LS means from the GEE model. This will be a secondary outcome for the VIONE trial when analyzed as a stand-alone trial.

Number of medication reviews completed by a pharmacist. This will be a secondary outcome for the VIONE trial when analyzed as a stand-alone trial.

This is a measure of count of medications used at the patient (not facility) level. This will be a secondary outcome for the VIONE trial when analyzed as a stand-alone trial.

High-risk DOAC use will be assessed by "flags" using the algorithm from an operations DOAC dashboard. These flags were developed based on existing guidelines and advice of many anticoagulation experts. Percentage of patients with PIMs across AD vs. LEAP+AD facilities were modelled longitudinally using Generalized Estimating Equations, with a three way-interaction of arm, month of follow-up, and pre-, post-period. Data will was be collected monthly at patient level and collapsed by clinic-month for patients who are at risk of potentially inappropriate medication use. Estimates computed using LS means from the GEE model. This is the primary outcome for the DOAC trial when analyzed as a stand-alone trial and the DOAC sub-analysis of the overall MIDAS study primary outcome.

This was the subset of patients with DOAC Population Health Management Tool (Dashboard) flags that occur when medications are given at doses that would be appropriate but are not because the patient has abnormal renal function. This has had minimal adjustments since being described in previous publications.

This was the subset of patients with DOAC Population Health Management Tool (Dashboard) flags that occur when medications are given at doses that would be appropriate but are not because the patient has unusually high or low weight or BMI. This has had minimal adjustments since being described in previous publications.

These are the remaining high-risk flags and are usually due to patients with medication interactions or doses that are incorrect due to the indication. This has had minimal adjustments since being described in previous publications.

Patient receipt of any CBTI will be measured by extracting from the medical records CBTI note templates completed by CBTI therapists. The denominator will consist of primary care patients who are not in hospice/palliative care. Percentage of PIMs across AD vs. LEAP+AD facilities were modelled longitudinally using Generalized Estimating Equations, with a three way-interaction of arm, month of follow-up, and pre-, post-period. Data will was be collected monthly at patient level and collapsed by clinic-month for patients who are at risk of potentially inappropriate medication use. Estimates computed using LS means from the GEE model. This will be the primary outcome for the CBTI trial when analyzed as a stand-alone trial.

Mean number of sessions will be measured by extracting from the medical records CBTI note templates completed by CBTI therapists. The denominator will consist of primary care patients who are not in hospice/palliative care. This will be a secondary outcome for the CBTI trial when analyzed as a stand-alone trial.

CBTI referrals will be measured according to counts of CBTI consult requests in the medical record. For clinics that do not use medical record consult requests specific to CBTI, referrals will be measured using monthly counts provided by CBTI therapists. The denominator will consist of primary care patients who are not in hospice/palliative care. This will be a secondary outcome for the CBTI trial when analyzed as a stand-alone trial.

The proportion of patients with a new prescription for a sleep medication for insomnia who have not had CBTI of those who are actively followed by the clinic and not in hospice/palliative care. This is a CBTI sub-analysis of the overall MIDAS study primary outcome.

3-item pilot measure of the extent to which employees engage in quality improvement activities given both at baseline and 18-months post-implementation. Scores are 1-5 with higher ratings indicating more engagement in quality improvement.

3-item measure comprising one item each for exhaustion, depersonalization, and reduced achievement (reverse scored) given both at baseline and 18-months post-implementation. "High Burnout" measures the percent of staff who are feeling burned out on all three burnout items at a frequency of "once a week" to "every day." Scored: 0-100%, where LOWER score is more favorable.

3-item scale. "Best Places to Work" (BPTW) is a summary measure of the group's satisfaction with the job, organization, and likelihood to recommend VA as a good place to work given both at baseline and 18-months post-implementation. The values are 1 to 5 with a higher score being more positive and the BPTW score is the average of the 3 questions. This is a measure normally administered within the All-employee Survey (AES) and the questions come from the Partnership for Public Service's BPTW survey (http://bestplacestowork.org).

7-item measure from the VA's newly developed Patient Safety Culture given both at baseline and 18-months post-implementation. Values 1 to 5 where higher values indicate more positive scores.