I have a question and the best way to describe is to use the draft I am writing.
See the draft:
Introduction
With the existence of federated repositories such as SHRINE, it is important to have reference ontologies for common data sources. In the past, the research community was able to suggest reference ontologies for diagnoses, procedures, medications and lab (core set of data feeds). Type of textual report filed in an EHR (e.g., spirometry report or radiology interpretation) can contain information which may complement the results of the current core set. For example an incomplete problem list may not mention asthma, but the existence of a spirometry report in an EHR (useful in assessing asthma severity determination) indicates that some diagnostic effort was performed for a given patient consistent with asthma problem hypothesis.
To be able to include types of diagnostic reports filed into our phenotype determinitation alghorithms for federated repositories, reference terminology for types of clinical reports is necessary. Our goal is to analyze suitability of LOINC to capture report types.
Methods
Review of the most frequent report types found in an IDR (integrated data repository) (or IDRs of multiple institutions) and determining whether a LOINC terminology has a concept for it.
Similar to [Lin, 2009, fall AMIA], and [vreeman 2007]
Marshfield Clinic IDR was used. Report titles which accounted for 80% of all IDR reports were chosen.
We used LOINC v 2.27 (Released 6/15/2009) was used in this evaluation
Results
At out institution: 35 report types account for 80% of all reports and 311 report types account for 99% of all reports.
(using a sample of 250k patients, lifetime data, 1960-2009)
We mapped those 36 reports to LOINC.
We found that LOINC does not have all the codes we need. (initial results indicate that about 30% of codes are missing) and many have some issues.
The LOINC codes seem to be predominantly from inpatient context.
Here are my questions:
1.
The DOC axis contains 529 concepts for reports types. How was this set created? When and how much evolution is there going on in the last year?
2.
Is there any effort within LOINC which looks at this currently?
Vojtech -
This sounds like a nice study. To answer your first question, let me point you to a couple of resources. This paper describes the origins of the initial Doc Ontology terms:
Frazier P, Rossi-Mori A, Dolin RH, Alschuler L, Huff SM. The creation of an ontology of clinical document names. Stud Health Technol Inform. 2001;84(Pt 1):94-8.
There is also a Clinical Documents tutorial available over on the LOINC website that gives some background, summarizes the current state and literature, and points toward a couple of future directions.
The revised set of document axis has been discussed for the past few years and was approved for use in the fall of 2007. The VA, Columbia, and others have been working on creating a new submission based on these values, but haven’t quite finalized them. In the meantime we have been slowly working on harmonizing the initial batch of LOINC Document Ontology terms that currently exist in LOINC with these new values. That is still a work in progress.
As for your second question, there are a couple of papers that have described the coverage of the LOINC ontology for various domains (nursing, hospital-based document titles) that are summarized in the tutorial. So, that might be a good place to start.
The document ontology work is a frequent agenda item at the Clinical LOINC Committee meetings, so would love to hear an update about your work and have you tune in to what the other folks have been up to.
Daniel,
Thank you for reply. I will try to follow up on this with the LOINC team.
Vojtech
I am pleased to report that Well Child Visit document type is now in LOINC !
Is there a list of how to interpret the source acronym.
for example what these mean:
PHS
UPMC
UPMC
DJV.MC
DJV.MC