Calendar

May 2018
M T W T F S S
« Mar    
 123456
78910111213
14151617181920
21222324252627
28293031  

Calendar

May 2018
M T W T F S S
« Mar    
 123456
78910111213
14151617181920
21222324252627
28293031  

Follow us on…

  • {name}
  • Google Profile
  • LinkedIn
  • Twitter
  • Facebook
  • YouTube
  • slideshare

Facebook

Recent videos

Posted on: February 22-2013 | By : Kunal Sawant | In: Healthcare Informatics | 6 Comments

While we talk about healthcare data standardization, it’s important to look at the constituents of healthcare data being captured within EMRs used by Providers. The following are the typical subsets of the information captured by providers within their EMR systems:

  • Patient Demographics
  • Lab orders & results
  • Medication orders
  • Medication administration records
  • Imaging orders & results (ECG, Echo, USG, CT, MRI, etc.)
  • Problems
  • Notes and documentation (SOAP, HPI, ROS, referrals, )
  • Physical examination
  • Diagnosis and coding (ICD, SNOMED, READ, LOINC)
  • ADT
  • Billing & Claims

 

Of the above, laboratory, medications and diagnosis data is the most structured since it does not include comments, descriptive notes or images that need interpretation.  Diagnosis data is captured by the ICD-9 and now the ICD-10 code sets, while Medication information could use RxNorm or the NDC code systems. For laboratory data, traditionally, most US care providers followed their own nomenclature for test names. This is, however, now changing with the ‘Meaningful Use’ mandates which recommend LOINC codes for describing laboratory test names within EMR systems. However, this entails a huge effort for mapping local provider laboratory test names to LOINC formats. To illustrate the challenge better, the following could be the local variations in the test names associated with the test Serum Glucose

 

TEST NAME (Glucose-related)

2HROB

DXGLUC

GLUC 28 WEEK

FS GLUC

GlucInOfc

GLUCLOSE

GLUHEM

 

Now,if were to analyzethe test names only and determine a match, it would be difficult to tag a correct standard LOINC test since these names do not exactly point to a correct test. To interpret, if these were really Serum glucose tests or even close to a glucose test, we possibly need some more insights into the test data. Let’s assume, you now have some additional information pertaining to these tests as shown below,

 

TEST NAME (Glucose-related)

UNITS

REF RANGE

2HROB

mg/dl

105-140

DXGLUC

mg/dl

85-125

GLUC 28 WEEK

mg/dl

100-130

FS GLUC

mg/dl

70-99

GlucInOfc

Null

Negative

GLUCLOSE

Null

Negative

GLUHEM

mg/dl

90-110

 

Mapping Laboratory Test NamesAs mentioned above, the units and reference range information gives us a better insight in understanding, interpreting and subsequent mapping of the above set of names. From our analysis of the presented data, most of the tests look to be tallying with the reference range and units associated with a serum glucose test. We can also see that test names ‘Gluclose’ and ‘GlucInOfc’ do not have numerical reference range or relevant units. These could possibly be Urinary glucose tests. If we had access to the actual observation values, it would enhance this analysis further so that we could confidently map these tests to a serum or a urinary glucose.

 

Thus, while mapping lab test names to a standard name, it is important for a terminology analyst to be presented data not just around the test name but also the reference range, units, observation or result values, the frequency distribution or spread of the observation results (incase these are numerical) as well as whether the results are numerical or textual. This would improve the accuracy of an analyst trying to map lab test data to a standard nomenclature such as the LOINC code-sets.

 

In our next post, we will explore and understand LOINC code sets and how to interpret laboratory test data to map it to these codes.

1 Star2 Stars3 Stars4 Stars5 Stars (2 votes, average: 5.00 out of 5)
Loading ... Loading ...

Kunal Sawant


Dr Kunalsen Sawant is medical physician with more than a decade of experience working with...

Read More...

 
 
  1. Jayant Singh on March 2, 2013 at 11:07 am Says:

    very interesting explaination to understand the problem and a basic mapping approach
    but actual analysis & implementation would be reall challange
    I am wondering
    1. if analyst has to do such analysis for each test
    2. chances of error & its consequences

    • Kunal Sawant on March 4, 2013 at 12:01 pm Says:

      Yes, its a complex effort with chances of errors. However, with the help of a trained terminology analyst and a tool presenting the analyst with the right information around the tests data, one should be able to get a better accuracy around the mapping.

  2. Kunal Sawant on March 8, 2013 at 10:27 am Says:

    Thanks for the comments Jayant. Regarding your queries,
    1. there are automapping tools which can map test names to standardized codes. But, the sheer variations in the test names used within hospitals, would need an element of manual intervention in handling the unmapped tests.
    2. To reduce errors while mapping, it is best that an analyst is presented not just the test name but also the underlying data sets pertaining to the reference ranges, units, % age of results which are textual v/s numerical so as to increase the chances of making a corrct map. A tool used for mapping these unmapped tests would best serve the purpose.

  3. Pingback: Eva

  4. Sugato Basu on June 20, 2013 at 12:47 pm Says:

    Does this mapper of Lab Data to std Nomenclature work for Payers who receive such data from 100s of Labs?

    • Kunal Sawant on June 20, 2013 at 6:44 pm Says:

      Yes, the terminology analyst can also work with nomenclature received from payers, provided he receives the lab data attributes. At a minimum the test name, the result values, reference range and units would be needed for an accurate map. You can also map using the test name only, but that leaves some scope for errors.

Leave a Reply

 

*

© 2018 Syntel, Inc.