Setting thresholds

Heather Leslie heather.leslie at
Sun Mar 4 20:21:33 EST 2018

Hi all,

Also consider differentiating between actual vs ideal ranges. All are intended to flag outliers via decision support and which may trigger clinical actions.

The reference range for a lab test result is very specific - at least for that lab, machine and methodology and in some cases it may be more targeted for gender and age etc. These are the actual ranges that apply for this test. It is common practice for the reference range to be added to an instance of data for a test result and this should be stored in the EHR alongside the actual result as the immutable means for the receiving clinician to interpret the result accurately. This is why there is a place in the RM for Ref Range.

An ideal of 120/80 for blood pressure is almost universally quoted. In the loosest sense this could be the generic ideal which could be used to trigger decision support. However it is not accurate in many situations where factors about the person and the context should override 120/80: a morbidly obese 95 year old man with multiple comorbidities and a starting BP of 160/110 might be aiming for something like 140/90 as ideal in the short to medium term; a very sick 2 day old baby who is 10 weeks premature in the neonatal ICU may require a very narrow range of acceptable BP measurements before decision support should trigger clinical action. These should be set as personalised targets by the treating clinician or may be drawn from an authoritative knowledgebase that may be external to the EHR itself.

An acceptable BMI will be vary due to combinations of the geographical location and gender, age and race of the patient.

None of the reference range values or recommended values should be added to the ADL in archetypes in CKM. At the moment we try to provide advice as to how the archetypes could be implemented within clinical systems and point toward external resources and knowledgebases where this is obvious and appropriate.

However, our desire to flag where measurements are outliers in clinical systems is reasonable and to be expected by end users. This is where clinical system vendors can differentiate - about how they apply intelligent algorithms to draw from clinical protocols, external knowledgebases maintained by experts or personalised goals and targets that may be defined by the clinician using the EVALUATION.goal archetype, which includes setting one or more targets that will contribute to achieving the goal. Bringing GDL into the CKM-type process could support this in the future, enabling clinical collaboration and verification or the algorithms and integration with external resources that could act as a source of truth for the variation in age, race, gender etc.

If we were able to include advice on external knowledge bases, protocols etc then this could be valuable but adds another layer of complexity in determining what sources should be recommended plus the time and resources required for governance and maintenance. In my experience, most projects have their own resources and sources of truth that they require to be included in a system and these could be conflicting with the 'best practice'.

It is a complicated area, indeed!



From: openEHR-clinical [mailto:openehr-clinical-bounces at] On Behalf Of GF
Sent: Friday, 2 March 2018 9:42 PM
To: For openEHR clinical discussions <openehr-clinical at>
Subject: Re: Setting thresholds

There are different kinds of ranges:
- normal with reference to a specific population,  for children, males, whites, etc
- signalling with reference to the specific subject of care
- signalling with reference to the specific healthcare provider
- signalling with reference to a specific context
- etc.

Gerard   Freriks
+31 620347088
  gfrer at<mailto:gfrer at>

Kattensingel  20
2801 CA Gouda
the Netherlands

On 1 Mar 2018, at 22:47, Sam Heard <sam.heard at<mailto:sam.heard at>> wrote:

HI All
Goals and targets are an example of ranges as data. The INR treatment range is a very common data set that the clinician populates as it is dependent on the history and other considerations.
Cheers, Sam

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the openEHR-clinical mailing list