Quantitative Data with Ranges

Owned by John Svirbely
Last updated: Apr 02, 2021
1 min read

Data, especially when quantitative in nature (e.g. labs, vitals), can be associated to “ranges”.
There are different types of ranges:

  • Possible/admissible ranges, determined by the nature of the entity being measured, and/or the measurement devices' precision.
    Example: Age ranges between 0 and ~125 years;
    the lower limit of most concentrations is technically 0, but “0” may not be viable, and instruments
    may not be able to distinguish between 0 and a concentration that is ‘too low’.

  • (Reference) Interpretation ranges, e.g. ‘normal’, vs ‘elevated’. These may be defined over a population,
    or fine-tuned for an individual.
    Range-based classifications are usually the output, rather than the input, of a decision/rule.
    See https://www.hl7.org/fhir/valueset-referencerange-meaning.html

  • Error/Confidence ranges: intervals associated to the instance data, usually to reflect some imprecision in the measurement process (e.g. 1000 +/- 10)

 

LOINC examples: laboratory test; heart rate; respiratory rate