Collaborative treatment

Partners HealthCare is one of the world’s preeminent institutions for the creation and application of medical knowledge. Partners, founded in 1994 by Brigham and Women’s Hospital and Massachusetts General Hospital, US, is an integrated, not-for-profit health system that includes primary care and specialty physicians, two academic medical centres, community and specialty hospitals, community health centres, home health and long-term care services, and other health-related entities.
In addition to its patient care mission, Partners is one of America’s highly regarded biomedical research organisations and a major teaching affiliate of Harvard Medical School. Both Massachusetts General and Brigham and Women’s are ranked in the top ten US hospitals by U.S. News and World Report.
While devoted to clinical and research excellence, Partners has also pursued the innovative use of information systems since its inception. Brigham and Women’s developed one of the first intelligent ‘computer physician order entry’ (CPOE) systems in the mid-1990s. Physician order entry systems now in use across Partners require physicians to enter all medications into a computer system at the patient’s bedside. The underlying clinical decision support (CDS) systems check to see if the orders are appropriate for the patient as well as consistent with commonly accepted best medical practices, and if not, alerts the physician that a different decision might be considered.
The physician can override the alert if he or she believes the order is the best treatment for the patient. These systems enable Partners to understand how clinician decision-making at the bedside lines up with commonly accepted clinical guidelines. Further, Partners monitors the override rate of alerts to determine if the knowledge base needs to be redesigned to be more effective and less task-interfering. This capture of physician decisions in the CPOE context is the first type of decision capture that Partners employs, and it is critical to the improvement of clinical processes.
Intelligent CPOE systems have been very successful at Partners and have been credited with a 55 percent reduction in medication errors. The systems have also allowed Partners to introduce new knowledge to the system – for example, new medication guidance based on genetic tests.
While the clinicians who make recommendations in the design of the CDS systems are highly-regarded (most are members of sanctioned committees such as the Partners Medication Knowledge Committee), there are circumstances for not following the system’s advice that only the patient’s physician can know. Partners’ executives and physicians are generally quite supportive of CPOE, and view it as a key component of a quality initiative called ‘High Performance Medicine.’
Outside of Partners, other hospital systems have adopted intelligent CPOE – some using home-grown systems like Partners’, and others employing increasingly capable packages from third-party vendors. National healthcare advocacy groups, such as the employer collaborative Leapfrog Group, argue that all healthcare providers should be using CPOE, and that medical errors would be dramatically reduced if they did. However, CPOE technology in itself is not sufficient to improve clinical practice and reduce medical error: some institutions have had organisational governance structures and cultures in which physicians largely rejected the use of CPOE.

The next frontier in medical decision making
Despite the success of CPOE at Partners, the institution is moving toward the next frontier of medical decision-making. In 2003 Partners established a formal (though small) knowledge management function to expand beyond the knowledge originally embedded in the CPOE system, and to manage knowledge across Partners in a more disciplined way.
The rules and other forms of knowledge embedded in the CPOE system were scattered across different parts of Partners, and processes were not well-defined or consistent for validating and updating these knowledge bases over time. The knowledge management function grew rapidly, and began to rationalise activities related to knowledge design. For example, it helped Partners establish business owners for critical enterprise knowledge domains, and established formal processes for update and maintenance. New areas of medical practice were also addressed, such as care protocols in such areas as cardiac disease, renal disease and diabetes. The protocols addressed treatment issues that went beyond medication, such as the importance of diet and smoking cessation in managing diabetic patients.
In order to successfully manage knowledge for healthcare, the managers of Partners’ knowledge management function concluded they had to begin a second form of decision capture. This form involves supporting and capturing an explicit decision-making process for knowledge base design in key areas of clinical care and medication management.
In many cases, it was difficult to go back and find out how and why a design decision was made which hampered the update process. While no malpractice or other legal case at Partners had ever involved looking at the process by which knowledge and decisions were formalised with decision support, it seemed possible that this could happen at some point in the future. The Partners knowledge management team believed similar design decisions are made with regard to medical device development, so they wrote policies and procedures for knowledge base maintenance modeled on the FDA medical device regulations.
Partners also implemented systems to capture discussions and decisions – either in votes or by acclamation – of clinical experts in an online forum. The tool used is eRoom from EMC Documentum. Each major clinical domain in which decisions are necessary has both a facilitator (a knowledge engineer from the Partners knowledge management organisation) and an expert review panel (the members of which are normally named by a high-level Clinical Content Committee at Partners). Each review panel has between four and ten expert members: over 200 experts from Partners participate in some online decision forum.
Discussion of some topics can range from days to months depending on the complexity, and may yield as many as fifty different postings as design decisions are resolved by acclamation or a vote taken. A decision discussion on a particular drug, for example, might address the most effective dose in an elderly or renally-impaired patient, the minimum and maximum doses for particular types of patients, when to interrupt and notify of a potentially unsafe order, and so forth. The knowledge engineer has the responsibility to guide expert panels efficiently as possible toward consensus.
The Partners knowledge management team focuses on clinical knowledge that reflects commonly accepted practices that are important to remember but easy to forget in the care process. Since the medical literature suggests some relatively simple aspects of treatment, such as medication dosing, are often wrong or inappropriate in clinical practice, patient care will benefit greatly from this straightforward approach. This strategy also facilitates rapid consensus development among expert clinicians.
More importantly, the Partners knowledge management team has determined these expert panels cannot efficiently make and capture design decisions without the assistance of a “decision steward,” in this case, the Partners knowledge engineer who is accountable for driving and capturing critical decision points in the design process.
After a set of decisions has been made in a particular clinical domain, the knowledge engineer formalises these decisions in a knowledge specification.
This version is only somewhat comprehensible by medical experts, but is primarily intended to drive the creation of logic embedded in the intelligent CPOE systems. As knowledge bases are updated either due to changes in clinical guidelines or as part of a scheduled maintenance review, the changes are validated with the expert panels in the eRoom environment, along with the rationale for the change in order to ensure ongoing capture of design decisions.
Over the three years that this decision capture approach has been in place, Partners has already seen maturation of the process. The governance structure for online content has been expanded and made somewhat formal. Knowledge engineers have become better facilitators, in part through coaching. Clinician experts have become more efficient in the use of their own time, and are more focused on the knowledge issues involved in making a design decision, rather than on purely intellectual topics. Participation by clinician experts remains high, even though it is largely voluntary. The Partners knowledge management organisation is applying the decision capture approach to new clinical domains, and is putting in place an overall knowledge architecture that would point to, if not fully integrate, key knowledge assets at Partners.

Decision capture outside partners
We believe that Partners’ approach to decision guidance and capture will eventually be found in many other knowledge – and decision-intensive environments. In some industries, such as pharmaceuticals, there are regulatory justifications for capturing decisions (and indeed, some firms in that industry are exploring the approach using some of the same technologies as Partners). In others, the rationale for capturing decision logic will be financial regulation or shareholder lawsuits.
Partners has been able to guide and capture decisions not only because of new technologies, but also because of an organisational structure and culture that encourages clinicians to participate in deciding how care decisions should be made and captured in systems. Other organisations will also find that they need both technology and organisation to succeed with decision capture. The technologies are commercially available, but the governance and cultural approaches that ensure accountable stewardship for decisions need to be home grown. Of the two capabilities, organisation is surely the more difficult.
Despite the difficulties of managing decisions in this way, there is little choice for some organisations about moving in this direction. What could be more important – in almost any industry – than how an organisation makes decisions? It probably won’t make sense for all decisions to be discussed online and captured, and it won’t be possible to develop systems like the CPOE to guide all daily operational decisions. For those decisions that are particularly important – whether to human life or to organisational survival – it is logical to begin applying technology and organisational structures to guide and capture decisions.

Tonya M. Hongsermeier, MD, is corporate manager, Clinical Knowledge Management and Decision Support, Partners HealthCare System, Inc., Boston, Massachusetts, US. Prior to joining Partners in 2003, she was VP for Knowledge Management Solutions and Patient Safety at CERNER Corp. Contact Tonya Hongsermeier at thongsermeier@partners.org to learn more.

Tom Davenport has been a long term observer of Partners. He holds the President’s Chair in Information Technology and Management at Babson College, where he also leads the Process Management and Working Knowledge Research Centres. His books and articles on business process reengineering, knowledge management, attention management, knowledge worker productivity, and analytical competition helped to establish each of those business ideas. His website is tomdavenport.com.