Kannayiram Alagiakrishnan, Patricia Wilson, Cheryl A Sadowski, Darryl Rolfson, Mark Ballermann, Allen Ausford, Karla Vermeer, Kunal Mohindra, Jacques Romney, Robert S Hayward

Elderly people (aged 65 years or more) are at increased risk of polypharmacy (five or more medications), inappropriate medication use, and associated increased health care costs. The use of clinical decision support (CDS) within an electronic medical record (EMR) could improve medication safety. Participatory action research methods were applied to preproduction design and development and postproduction optimization of an EMR-embedded CDS implementation of the Beers' Criteria for medication management and the Cockcroft-Gault formula for estimating glomerular filtration rates (GFR). The "Seniors Medication Alert and Review Technologies" (SMART) intervention was used in primary care and geriatrics specialty clinics. Passive (chart messages) and active (order-entry alerts) prompts exposed potentially inappropriate medications, decreased GFR, and the possible need for medication adjustments. Physician reactions were assessed using surveys, EMR simulations, focus groups, and semi-structured interviews. EMR audit data were used to identify eligible patient encounters, the frequency of CDS events, how alerts were managed, and when evidence links were followed. Analysis of subjective data revealed that most clinicians agreed that CDS appeared at appropriate times during patient care. Although managing alerts incurred a modest time burden, most also agreed that workflow was not disrupted. Prevalent concerns related to clinician accountability and potential liability. Approximately 36% of eligible encounters triggered at least one SMART alert, with GFR alert, and most frequent medication warnings were with hypnotics and anticholinergics. Approximately 25% of alerts were overridden and ~15% elicited an evidence check. While most SMART alerts validated clinician choices, they were received as valuable reminders for evidence-informed care and education. Data from this study may aid other attempts to implement Beers' Criteria in ambulatory care EMRs.