Health Information Technology Center for Education and Research on Therapeutics

The Centers for Education and Research on Therapeutics (CERTs) demonstration program is a national initiative to conduct research and provide education that advances the optimal use of therapeutics including drugs, medical devices, and biological products. The purpose of this health information technology (IT) CERT was to leverage health IT for three independent research projects focused on: 1) new technologies to improve pharmacosurveillance, 2) using new sources of data from clinical decision support (CDS) to identify physician-level variation, and 3) examining human factors principles in the design and implementation of medication-related CDS.

The specific aims of this project were as follows:

e-Pharmacovigilance II: Surveillance for Safety and Effectiveness

• Develop a patient-reported, electronic health record (EHR)-integrated system to actively monitor the safety and effectiveness of treatment for patients taking Food and Drug Administration-approved medications for one of four common chronic conditions (diabetes, hypertension, insomnia, depression), with integrated management support by a pharmacist. 
• Measure the reach, effectiveness, adoption, and implementation of this integrated module for adult primary care patients in the Brigham and Women's Primary Care Practice-Based Research Network. 

Physician-Level Variation in Medication Overrides of CDS 

• Evaluate how much clustering of overrides there is by provider for both safety- and efficiency-related issues in the inpatient and outpatient settings. 
• Evaluate the appropriateness of overrides overall and among those with high override rates. 
• Characterize and understand the patterns and reasons for these override behaviors and decisions. 
• Intervene with providers who have high override rates, for the specific categories that appear inappropriate. 

Examining Human Factors Principles in the Design and Implementation of Medication-Related Decision Support Alerts

• Validate the human factors instrument on vendor and in-house medication-related decision support systems and determine how a "good" score on Instrument 1 correlates with override rates. 
• Determine whether employing human factors principles can have an impact on physician behavior in terms of the rates of overriding alerts.
• Evaluate user-related outcomes for systems that comply with human factors principles. 

The CERT established two cores that supported the three projects. The first core was the methodology and data resources core, which focused on coordinating study design and analytic strategies across research projects. The second, the translational and dissemination core, promoted interchange and cross-fertilization between projects, dissemination of research findings, and the translation of findings into practice through initiatives at the local, regional, and national levels.

Key findings are presented by project:

The e-Pharmacovigilance project conducted a cluster randomized trial at 18 clinics to test the efficacy of an EHR-linked, interactive voice response (IVR) tool that provided the support of a live pharmacist to detect patient-reported adverse drug symptoms. More than a third of patients reported symptoms and were transferred to the pharmacist, who assessed the severity and likelihood of symptoms being related to the medication. Chart reviews are in progress to identify the total number of physician-noted symptoms related to target medications, and to examine hypothesis that the IVR calls and pharmacist intervention would significantly increase the numbers of medication-related symptoms identified by physicians.

The Physician-Level Variation in Medication Overrides of CDS project evaluated the frequency and appropriateness with which physicians overrode alerts and the override reasons provided. The results indicated that more than half of the medication CDS alerts were overridden, with reasons varying depending on the type of alert. Many alerts that were overridden inappropriately had the potential to cause patient harm. The research team plans to use data collected from the project to identify physician-level variations and use the results to improve patient safety and efficiency.

The Human Factors Principles in Design and Implementation project used surveys and observational scenario-based analyses of clinical reasoning to assess how the design of alerts affects prescribing behavior impacting CDS acceptance and use. The study found a negative correlation between alert design and alert override rates, where alerts rated as better designs were less likely to be overridden. The survey showed that high alert volume was associated with higher likelihood of participants overriding alerts and not paying attention to even clinically significant alerts. Declining an alert suggestion was preceded by sometimes brief but often complex reasoning, prioritizing different aspects of care quality and safety, especially when the perceived risk was higher.