Making a case for clinical decision support
Two-thirds of hospitals have clinical decision support mechanisms in place, but many are vastly underleveraged, according to HIMSS Analytics.
During Monday’s session “Managing Clinical Decision Support Knowledge at the Speed of Change,” Tonya Hongsermeier, MD, MBA, corporate manager clinical decision support and knowledge management, at Partners Healthcare System, Inc. in Boston, outlined a business case for investing in knowledge management and shared a successful model of governance and infrastructure for knowledge management.
According to Hongsermeier, multiple forces drive the need for decision support and knowledge management. For starters, the volume and velocity of knowledge processing continues to grow; currently physicians need 2 million facts to practice medicine. Over the next decade, the issue will be exacerbated by the increasing availability and utilization of molecular diagnostic tests, said Hongsermeier, which will increase the speed of change of evidence exponentially.
Clinical knowledge needs are not the only decision-support driver, says Hongsermeier. Pay for performance incentives are emerging on the market and causing health systems to look at EHR and decision support mechanisms to improve performance.
Hongsermeier also outlined some challenges with vendor decision-support tools. They tend to interrupt tasks and workflow rather than fitting into workflow. In addition, knowledge management editors “silo-ize” and segregate content without support for long-term issues and challenges. Finally, most lack support for front-end decision-making. That is, systems tend to be responsive rather than anticipatory.
Decision-support mechanisms should integrate into physician workflow, consolidate and link data and facilitate rapid changes in order sets and rules when the formulary changes, said Hongsermeier.

Partners HealthCare, established among major Boston hospitals in 1994, has grown to thousands of community physicians. The organization has a long history in applied clinical decision support including inpatient interactive order rules and alerts. Hongsermeier shared the decision support implementation process created and used at Partners Healthcare to illustrate a working model for multi-enterprise decision support.
Implementation steps are:
  • Identify stakeholders and determine goals;
  • Evaluate vendor and internally-developed solutions to determine what can be accomplished;
  • Build teams to design decision support interventions;
  • Develop and pilot an implementation plan;
  • Implement the plan across the healthcare system; and
  • Measure objectives and create a maintenance plan.
The Partners process started with knowledge engineering teams, who provided an overview of key challenges: content ownership questions and disputes, IT/quality management alignment and inadequate editors.   

The team also calculated the costs of knowledge engineering life cycles to determine the cost of not having decision support and understand the benefits associated with an increase in the speed of translation of evidence into clinical decision support. Partners also identified other value propositions like stakeholder involvement in the design process to support EMR adoption, avoidance of potential liability and alignment of knowledge assets with quality and safety requirements.
Partners’ three-stage decision support upgrade began in 2004 and focused on transparency and governance, collaboration and life cycle tools and content editing re-architecture. At the governance stage, Partners created clinical content committees to oversee and update content.
The group relied on a strategic knowledge worksheet mechanism for collecting data and content needs to support appropriate decision-making. Partners’ knowledge management portal allows key word searching, drilling at the taxonomy level and filter-based searching
The second phase, the content management collaboration phase, supports content lifecycle management tools to reduce or eliminate delayed decisions and lost knowledge. Partners implemented “eRooms” organized around governance structures and priorities. All decision-making assets are pushed through eRooms to manage implementations and updates.
This phase also created a system to track and resolve disputes and other issues. For example, editing evidence into decision support may require the site to define diabetics with renal disease and hypertension and propagate data in multiple directions to define a single medication contraindication. Partners created SmartForms to support guided ordering, consistent information capture and decision making and help aggregate definitions and “de-silo-ize” content.
Hongsermeier shared several key questions to steer the project and create a foundation for success.
  • Who is accountable for content?
  • Are assets on an update schedule?
  • How long does it take to design content?
  • Finally, is decision support changing business performance and are clinicians/end-users satisfied?
Hongsermeier concluded with a final piece of advice for healthcare enterprises eying decision support tools. “Vendors offer excellent decision support resources. It’s better not to build what can be bought, but remember all solutions will require internal resources to customize and meet site-specific needs.”