I have noticed that some applications generate Result Observation entries where the value type doesn't match the LOINC code scale type. For example, I have seen observations for LOINC term 4548-4 (Hemoglobin A1c) where the <value> element has an xsi:type="ST" attribute even though the LOINC scale part is Quantitative (Qn) and the actual value is a number.

Add a new bullet point under CONF:1198-7143 as follows.

• If code contains a valid non-null LOINC then the xsi:type value SHOULD match the LOINC Scale Part based on the table below.

I recommended making this new conformance rule a SHOULD rather than a SHALL because some applications such as EHRs are at the mercy of upstream data sources which don't always use the correct data type. For example, a Laboratory Information System (LIS) might have reported an out-of-range Hemoglobin A1c observation as an ST data type with a value of ">14.0" rather than the correct approach of using an IVL_PQ data type.

For precedent please see the HL7 Version 2.5.1 Implementation Guide: Laboratory Results Interface Release 1, STU Release 3 – US Realm, Table 8-18 (page 191). It gives an explicit mapping from LOINC scale types to allowed HL7 V2 data types. We should do the same in C-CDA 2.1 for HL7 V3 data types.