The primary characteristic of a CDA document is that it must be readable by humans. The CDA specification states that, “human readability guarantees that a receiver of a CDA document can algorithmically display the clinical content of the note on a standard Web browser.” This requirement means that a clinician or patient can take a CDA document and display it on any computer with a web browser without the need to load any additional application.
The content of the CDA document displayed by the web browser is contained in a portion of the document called the CDA Narrative Block. The CDA Narrative Block contains the textual representation of the clinical event being described.
As shown in Figure 1, immediately after the CDA Narrative Block, the CDA document may also include optional coded entries that represent the clinical event using common coding systems such as LOINC, SNOMED, and others.
There are two ways to populate the CDA Narrative Block:
- Directly Authoring. This involves populating the CDA Narrative Block with textual comments provided directly from the physician. The comments could come from dictation or the writing or typing of notes.
- Derivation from Coded Entries. The text in this case is derived from coded entries. The physician would enter information into an EMR in a point-and-click manner, which allows the clinical event to be documented using coded entries. When a CDA is exported by the EMR, the CDA Narrative Block is machine generated from the coded entries.
There are advantages and disadvantages for each method of populating the CDA Narrative Block. In the use case of CCD, the derivation of the text is the recommended approach. This allows for more advanced machine-to-machine sharing of data, assuming the receiving system parses out the coded entries and integrates them with the local patient health information.
Pros and cons of directly authoring the text:
- Thought processes are sometime better represented by open narrative. The complex story of the patient can be difficult to express by point-and-click.
- Directly authoring requires more manual intervention such as dictation processing, hand writing, or typing of information. Natural language processing (NLP) can aid in this effort.
- It is difficult to ensure that optional coded entries match the narrative text. More advancement with NLP could help with this as well.
Trade-offs for deriving the text from coded entries:
- Aids in clinical decision support tools, reporting, and quality analysis.
- Ensures that the coded entries should match the narrative text that is displayed via a web browser.
- Using point-and-click entry forces the physician to enter information in a structured way, which may cause some semantic meaning to be lost. Point-and-click may also slow down the treatment process until workflow adjustments are made.
- Different systems may utilize different coding systems, which can add difficulty to the machine-to-machine sharing of data.
The mere fact that these trade-offs are now being analyzed means the industry is heading in the correct direction. The first step with Meaningful Use Stage 1 was to prompt the industry to exchange clinical health information at a basic level. I believe that has been a success.
The next step is to work through the details of how to make the data more meaningful and interoperable. Simply narrowing the list of utilized coding systems, as discussed in my previous blog, can have a large impact on making the clinical data more interoperable. Another sign that the industry is headed in the correct direction.
These issues will be sorted out as more adoption and use take place. Whether we invent more powerful tools that convert human language to specific codes or discover faster ways for doctors to enter structured data that maintains semantic meaning, the first step of utilizing CDA documents will naturally push the industry to more coordinated and better quality of care.