E-Health: LOINC standard

Holger · February 22, 2017, 11:27am

Primer: E-Health is the efford to digitalize the health care system. This will allow to exchange all the important medical data from one medical facility to the other. It will also allow to export your device data in a way that is compatible with a wide range of digital services (pump data, sensor data, meter data, logs). The relevant standards are HL7, LOINC, UCUM and others. This discussion is part of a series that is aiming at raising awareness for E-Health related issues. Important future decisions are now made on our behalf. Our voices as patients should be heard by manufacturers and government organizations. This might be a bit technical but the more attention this is getting the more momentum it will have.

Goal: we all know the web and the important thing about the web is that it works according to agreed standards like HTML, XML and CSS. These standards have been developed by the World Wide Web Consortium (W3C). It does not matter which operating system you are using. The only prerequisite is that your browser is working according to the W3C standards. For the exchange of medical data we want the same. No matter what device you are using the data exchange should follow international standards. Imagine that you could export your device, app or cloud data and your doctor could process this data easily. It would be a great relief for you and your doctor because you both could use the software, devices and services you prefer and know best (like Firefox VS Chrome). Here the standard Health Level 7 (HL7) describes the general XML structure of the document.

Discussion: the abbrevation LOINC is standing for Logical Observation Identifiers Names and Codes. This is a standard to describe the characteristics of a data field in the XML structure:

<result>
  <name>
    <coding>
      <code>14749-6</code>
      <system>http://loinc.org</system>
      <display>Glucose</display>
    </coding>
  </name>
  <valueQuantity>
    <value>123</value>
    <units>mg/dL</units>
  </valueQuantity>
</result>

This describes one value of 123 measured in the scale mg/dL. The important part is the coding. Here the LOINC code 14749-6 is used. This code stands for Glucose measured in blood plasma.

I have been reading through all the Glucose related LOINC codes. What I am missing is a LOINC code for Glucose measured in the interstitial fluid. This is differentiation between plasma and interstitial is important for Glucose sensors. Therefore I have asked in the LOINC forum for a new LOINC code to allow this differentiation.

I am also missing LOINC codes to characterize insulin. The standard might be able to express that Humulin has been injected for the purpose of eating 40g of carbs:

<result>
  <name>
    <coding>
      <code>14796-7</code>
      <system>http://loinc.org</system>
      <display>Humulin</display>
    </coding>
  </name>
  <valueQuantity>
    <value>8.5</value>
    <units>IE</units>
  </valueQuantity>
</result>

<result>
<name>
  <coding>
   <code>75296-4</code>
   <system>http://loinc.org</system>
   <display>Carbohydrates</display>
  </coding>
</name>
<valueQuantity>
  <value>40</value>
  <units>g</units>
</valueQuantity>
</result>

I was only able to find the Insulin Humulin in the LOINC codes. By just introducing Humulin with a LOINC code this would make it necessary to introduce all the other brands as well. In my opinion it would be more helpful to have a LOINC code to express what the insulin was used for like BASAL or for BOLUS FOR CARBS or for BOLUS FOR CORRECTIONS. In that case an additional coding could be used to express the brand name itself.

What do you think? Introduce all brands as individual LOINC codes or focus on the purpose of the insulin injection?

Bowie · February 22, 2017, 10:11pm

Sounds like the logical path to me otherwise it obsoletes itself when the better than humalog treatments is developed. What’s taking so long for the global medical field to standardize anyway!

Thas · February 22, 2017, 10:40pm

Though I never studied the details of the various coding systems, the conversion to “E-Health” systems and the integration of the many standards and non-standard systems were key issues that were focused on in my study of Healthcare Informatics (for which I hold a Masters Degree – I know, useless information ).

i think that despite the added work necessary, the more detailed the coding, the better. Lack of detail leads to all sorts of mistakes or coverage issues (for example). So yes, introduce MORE detail for better standards.

truenorth · February 22, 2017, 11:06pm

This isn’t my area of expertise but I’d think having more inclusivity would foster better long-term use and value.
Here in the US where the bureaucratic machine of “insurance” is apparently bigger and weirder than anywhere else in the developed world, there is a code for everything (and heaven help you if someone accidently enters an incorrect code).
Adding a few codes for the couple dozen types of insulin in use seems trivial by comparison, even if you include such differences as blood or interstitial bg values, purpose of use and so on.
Appreciate your efforts to get better data more easily into the hands of health professionals.

Holger · February 23, 2017, 10:45am

Your reply is why I have posted this here. I really hoped to get the feedback of someone with your professional background!

Imagine we propose to the LOINC people that we need LOINC codes for every brand of insulin. With this path we can express the exact insulin that the patient is using. At the same time we loose the information what the insulin was used for. The differentiation between “Bolus for Carbs” and “Bolus for Corrections” would be lost for example. But this differentiation in the diabetes log is telling a lot about the capability of the patient to manage his diabetes. For people reading diabetes logs professionally like endos and diabetologists this might be insufficient.

The HL7 Standard allows multiple codes. Therefore I propose to use LOINC codes for the brands AND for the purpose of the injection. Here is an example the patient had 200 mg/dl and applied 2 IE of Apidra for Bolus Correction. At the same time the patient will eat 40g of carbs. For this he injects 8 IE as the Bolus for Carbohydrates.

<result>
<name>
  <coding>
   <code>14749-6</code>
   <system>http://loinc.org</system>
   <display>Glucose</display>
  </coding>
</name>
<valueQuantity>
  <value>200</value>
  <units>mg/dL</units>
</valueQuantity>
</result>

<result>
<name>
  <coding>
   <code>14749-701</code>
   <system>http://loinc.org</system>
   <display>Apidra</display>
  </coding>
  <coding>
   <code>50000-2</code>
   <system>http://loinc.org</system>
   <display>Bolus Correction</display>
  </coding>
</name>
<valueQuantity>
  <value>2</value>
  <units>IE</units>
</valueQuantity>
</result>

<result>
<name>
  <coding>
   <code>75296-4</code>
   <system>http://loinc.org</system>
   <display>Carbohydrates</display>
  </coding>
</name>
<valueQuantity>
  <value>40</value>
  <units>g</units>
</valueQuantity>
</result>

<result>
<name>
  <coding>
   <code></code>
   <system>http://loinc.org</system>
   <display>Apidra</display>
  </coding>
  <coding>
   <code>50000-1</code>
   <system>http://loinc.org</system>
   <display>Bolus Carbohydrates</display>
  </coding>
</name>
<valueQuantity>
  <value>8</value>
  <units>IE</units>
</valueQuantity>
</result>

Here we need addional LOINC codes to be able to express the “injection-meal interval” (high BG waiting for the insulin to unfold its activity) in minutes and “meal-injection interval” (low BG waiting for the carbs to be digested and then injecting the insulin).

Would you agree?

Thas · February 23, 2017, 2:32pm

I agree. If we’re tying to record this information in a usable form, we need ALL of the pertinent information; otherwise we risk making poor decisions based on incomplete data.