Not able to represent decimals with higher precision than Float

[Basilio Bogado]

Describe the bug

We're mapping an OBX.5 Observation Value. It is a decimal value that requires lossless precision when converting to FHIR. The value we have in HL7 is 1769.859285 and it gets rounded to 1769.8593 when transformed to FHIR. We need to preserve decimal precision without rounding when transformed.

We found out that the rounding is happening because hl7v2-fhir-converter is using Java's Float when mapping OBX.5, which is single-precision and represents a very limited range of values.

The problem is, AFAIK there is no support for a decimal type with higher precision (like Double or BigDecimal)

To Reproduce

We're trying to convert the following OBX segment in an ORM_O01 message:
OBX|1|NM|8339-4^BIRTH WEIGHT MEASURED^LN|1|1769.859285|g^gram^UCUM|||||O|||20230506050000-0500|||||||||||||||QST|AOE

And we get the following value as a result:

"valueQuantity" : {
  "extension" : [ {
    "url" : "https://reportstream.cdc.gov/fhir/StructureDefinition/alternate-value",
    "valueString" : "1769.859285"
  } ],
  "value" : 1769.8593,
  "unit" : "g",
  "system" : "http://unitsofmeasure.org",
  "code" : "g"
}

Expected behavior

We expect the conversion to return "value" : 1769.859285 instead of "value" : 1769.8593

The type for the value is defined in Quantity.yml as follows:

value_1 :
  condition: $value CONTAINS_STRING '.'
  type: FLOAT
  valueOf: $value

We'd like to be able to have something like:

value_1 :
  condition: $value CONTAINS_STRING '.'
  type: DECIMAL
  valueOf: $value

Where DECIMAL would map to BigDecimal, which would preserve the value without rounding. Double could also work

Desktop (please complete the following information):

• OS: macOS
• Version: Ventura 13.4.1

Additional context

We can volunteer to create a PR to add an additional decimal type if that makes sense

[Lisa Wellman]

In release 1.1.0