Saturday, March 8, 2014

Modeling Care Pathways with BPMN

The optimization of healthcare delivery can be achieved through the standardization of care pathways and treatment protocols based on the latest scientific evidence. The following are some of the benefits of the standardization of care pathways:
  • Improvement in patient outcomes in the context of a shift from a fee-for-service to a value-based healthcare delivery model
  • Reduction in the variability in care quality
  • Facilitating the dissemination and adoption of evidence-based practice (EBP).

Different formalisms have been proposed over the years for representing the medical knowledge in clinical guidelines. Examples include GLIF (Guideline Interchange Format), Asbru, and PROforma. However, these formalisms have been largely confined to academia.

In this post, I discuss the use of the Business Process Modeling Notation (BPMN) for modeling and representing existing Clinical Practice Guidelines (CPGs) and Care Pathways (CPs). Once represented in BPMN, these guidelines can be translated into computer executable guidelines in the form of Clinical Decision Support (CDS).

As an example, let's review the algorithm below titled Management of substance use disorder, Module C: General Health Care which has been extracted from the VA/DoD Clinical Practice Guideline (CPG) for the treatment of substance use disorders.

The advantage of using BPMN is that it is widely used across industries and implemented by several commercial and open source tools. The algorithm above is not represented in BPMN notation. However, it is relatively easy to create a BPMN representation of the same diagram and doing so can actually help elaborate and clarify its use within a specific healthcare organization.

Clinicians often complain that CDS systems are not well integrated with clinical workflows. One benefit of using BPMN for modeling CPGs is that it facilitates the integration of business rules and business processes. As can be seen from the algorithm above, CPGs are typically represented as a network of tasks some of which are decision tasks. For example, in the diagram, there is a decision node with the question: Are Treatment goals achieved?. This decision task can be modeled in BPMN as a business rule task and executed by a business rule engine at run time.  There are business process management (BPM) and business rule management systems (BRMS) tools that provide this integration out-of-the box.

At a high level, BPMN represents business processes using events, activities, data objects, gateways, tasks, and grouping elements. Gateways control the splitting and merging  of sequence flows in a business process. The different types of gateways include: Parallel gateway, Exclusive gateway, Inclusive gateway, and Complex gateway. Gateways can also be implemented by a business rule engine. Data objects allow interaction with clinical data sources such as as Electronic Medical Record (EMR) systems.

Grouping elements like pools and lanes are used to delineate the responsibilities and roles of individual clinicians or organization in the process. For example, while the algorithm above is for the management of substance use disorder in general (primary) care, there is a note specifying that patients may be offered referral to addiction  specialty care at anytime. This separation of roles can be captured with lanes.

In addition to business rule tasks, BPMN also supports abstract tasks, service task, user task, and script tasks. A user task represents a human task and can support the creation of task-oriented user interfaces in CDS systems. A separate specification called the Web Services Human Task (WS-HT) specification provides an application programming interface (API) and a coordination protocol for interacting with human tasks in a service-oriented manner. The state diagram for human tasks below extracted from the WS-HT specification shows the different states of a human task and the transitions between them. The WS-HT also support notifications which can be used for alerts and reminders in clinical decision support.

No comments:

Post a Comment