- Tim O'Connell - CEO emtelligent
Continuity of Care Document
‘Continuity of care document’ (CCD) is a generic term for a type of computer-generated, patient-specific document that contains information about that patient’s health. Present-day CCDs are based on the HL7 Consolidated Clinical Data Architecture (C-CDA), which is an attempt to create a standards-based means of patient information exchange.
While the name ‘continuity of care document’ implies that CCDs are created during transitions in care (such as at time of hospital discharge or transfer), CCDs are being commonly used for other purposes, such as in the exchange of patient information between a provider and a payer.
Even though they are standards-based, CCDs can vary widely in their content depending on the software that is used to generate them. Stage 1 of the Meaningful Use program from the Centers for Medicare & Medicaid Services (CMS) originally specified that CCDs must contain a problems, allergies, medications and lab results sections; Stage 2 required additional data, including vital signs, smoking status, care plans and more.
A CCD file is a necessarily complex document, formatted in the Extensible Markup Language (XML), and containing multiple sections that contain both HTML-formatted and XML-formatted data (figure 1, at left). The HTML data and XML data are linked by unique identifiers, and there can be slightly different data in both sections.
Depending on how the electronic health record (EHR) system is configured to create CCDs, they may be very large, containing numerous historical records, or they may be abbreviated, for example just containing the most recent patient information.
While CCDs always contain list-form structured data, they also frequently contain free-text embedded documents and free-text notes, such as in their ‘Results’, ‘Plan of Care’, and ‘Interventions Provided’ sections. Depending on the patient’s history and how the CCD was exported, tens to hundreds of these reports can be contained in a CCD; with kilobytes to megabytes of free-text data. Additionally, free-text can appear as a ‘Comment’ relating to structured data in nearly any other section as well.
Examples of embedded documents might be an operative (surgical) report, diagnostic reports such as radiology, pathology, gastroenterology, or cardiology reports, the full text of which can be embedded in the ‘Results’ section of a CCD. See example of an Endoscopy Report embedded in a CCD below:
Unstructured Data in CCDs
The free-text clinical notes and reports contained in CCDs are extremely valuable documents for care providers, payers, and other CCD consumers, as these often contain specific procedure details, diagnostic data, or other results that are pertinent to patient care. An example could be the operative report from a patient’s prostatectomy detaling an operative complication such as bladder or nerve injury, or a pathology report from the same procedure containing a Gleason score from the patient’s prostate cancer. These are obviously valuable pieces of information that may not be captured elsewhere in the CCD, and can be used to predict future morbidity for the patient.
While emtelligent’s emtelliPro® medical NLP engine converts medical free-text into usable structured data, and can be used to unlock the value contained in these documents, emtelligent has gone a step further, and created unique software features that allow the linking of both the unstructured and structured data in CCDs. An example of this would be associating structured document metadata such as report type and document date along with the structured data extracted from free-text within these reports; an example data table of this type of data is shown below:
Left Ventricular Ejection Fraction
LVEF is visually estimated at 35%.
Gleason score of 7.
Forced Expiratory Volume in 1 second
FEV1 was 68% predicted on PFTs performed today.
Left Ventricular Ejection Fraction
Left Ventricular Ejection Fraction was 20%.
The value of this is immediately clear - CCD consumers can have access to this valuable data, and view it in temporal order to track progression of health and disease, linking it back to the source document so that data providence can be assured.
emtelliPro & emtelligent Background
emtelliPro is emtelligent’s highly-accurate, deep-learning-based medical NLP engine. It is scalable from a single server to hundreds of servers, capable of processing millions of documents per day. Created by emtelligent’s team of NLP experts and physicians, it uses the latest technologies to ensure industry-leading data extraction.