The following is a guest post by Adam Lokeh, M.D., vice president of clinical development and informatics with Wolters Kluwer Health.
Clinical decision support tools (CDS) play an increasingly critical role in a healthcare organization’s overarching strategy to comply with federal incentive programs and succeed within the quality- and performance-based reimbursement landscape currently unfolding. When effectively aligned with physician documentation practices at the point of care, these tools can have a powerful impact on error reduction, the standardization of evidence-based practices, quality of care and ultimately saving lives.
Research reveals that a combination of advanced CDS technology working in tandem with computerized physician order entry (CPOE) solutions will be needed to successfully navigate the coming healthcare landscape. A number of CDS elements will need to be considered and integrated into existing systems to create this powerful collaboration including evidence-based order sets, alerting systems for medication management, ECA rules (event, condition, action), referential information including guidelines and care plans, smart documentation and surveillance technology to name a few. To fully leverage the advantages of these tools, it’s important to understand the different approaches to data and content and the inherent advantages and disadvantage of each.
Currently, there are two approaches to content when designing point-of-care IT infrastructures—structured and unstructured. While both have the potential to standardize care and improve decision-making, industry preference leans toward greater integration and use of structured content for its ability to lay a foundation of improved accuracy, efficiency and ability to drive clinical decision support and analytics.
Because structured content is tagged or coded data that resides in a fixed field, it can be easily located, identified and understood, simplifying the process of integrating content into existing systems and sharing between disparate systems. In contrast, unstructured content, such as free text, often results in irregularities and ambiguities that make it harder to interpret.
Unstructured data makes it more difficult for health IT systems to recognize shared data, requiring complex and largely manual conversion processes that are prone to errors, resulting in inaccurate data. When inaccurate patient information is then shared between systems, the potential for adverse events or care issues is only compounded.
While the premise of this discussion as it relates to the benefits of structured content would seem clear, it’s not that simple. Physicians want the ability to express themselves freely when documenting, and there is concern within the physician community that the full patient picture could get lost if the narrative is too highly structured. As a practicing physician, I understand the delicate balance that exists between the need for a technological foundation that promotes accurate information sharing and the desire to protect the individual patient story.
The truth is that there can be risk without allowing for flexibility in creation of narrative content. Poorly-designed interfaces have clearly existed with some structured content frameworks historically—and still do today within some CDS and CPOE applications—that can cause pieces of the patient narrative to get lost. The use of applications lacking in flexible design and without trustworthy content that is thorough and exhaustive in nature has led to poor physician perception and even fear that the technology will marginalize patient care. Ultimately, the end result is poor physician adoption.
That is why it is so critical that vendors work with physicians to identify all essential elements as well as the factors that can hinder adoption. The solution is new, thoughtful clinician-designed systems that are more intuitive and flexible, allowing some limited unstructured content to help flesh out the narrative.
When CDS technology is developed through this kind of high-level partnership and designed to accommodate the use of structured content where it is needed most, content can be indexed at a granular level, easing the process of mapping within systems. It also lays a foundation for automated updating of content as industry evidence changes and provides a framework for more robust reporting due to extensive filtering capabilities.
The end result is more accurate and efficient integration of the best industry evidence at the point of care, delivering a framework for decision support that truly impacts care without compromising the patient narrative. It’s this kind of far-reaching potential—currently offered through some of the more advanced CDS and CPOE applications in the industry—that physicians need to witness to truly understand what can be accomplished. Unfortunately, the industry has not done a very good job of educating them to date.
Some are looking to the potential of natural language processing (NLP) to address the needs for mapping in free-text environments through data mining. While this path offers an alternative, it is not as powerful a foundation as structured content for improving decision making at the point of care. In fact, it’s retroactive. If data mining occurs after the patient narrative has already been input, decision support can, by definition, only be offered “after the fact.”
In essence, physician documentation that is completed in a structured-content environment —as opposed to a traditional dictation method—is, in itself, a form of CDS. Because documentation can be structured to guide and remind physicians to document important medical elements, it assures that nothing is overlooked.
Many industry initiatives point to greater incorporation of structured content into the design of IT applications for information exchange. Industry movements and organizations such as Meaningful Use, HL7, the Standards and Interoperability (S&I) Framework Health eDecisions Project and the CDS Consortium are working towards industry standards that will require use of more structured content.
The simple fact is that when data is shared, it has to be recognized across and between systems. Structured content within CDS applications allows data to be mapped to a standardized vocabulary to ensure accuracy.
That said, clinicians prefer free text. Until the industry properly educates physicians regarding the power inherent in structured content, the best approach will be a hybrid that includes avenues for both models. For maximum adoption, IT vendors should consider that critical components will need to be structured to drive CDS, reporting and quality metrics, but allowing for some amount of free text to smooth out the edges for more widespread adoption.