1. It improves interoperability—I don’t agree that this is always the case. Unnecessary tightening of constrains in a spec reduces or eliminates the spec’s re-usability as the parent of other IGs and curtails interoperability. The result is that the templates are loaded with so many required elements that they only serve the narrow use case conceived at the time of writing.
2. It ensures data uniformity—While true that an IG with SHALL bindings everywhere will result in rather uniform CDA documents, I disagree that the benefits of uniformity outweigh the good of having a flexible spec.
3. It encourages data capture—This is one of the most common but spurious ideas regarding IG design. Expressed in several ways it states that “if a data element is bound by a SHALL verb, the clinicians and vendors will begin to collect and record the data.” This may be true some of the time but more often leads to nulling. The SMART C-CDA Collaborative did an analysis of CDA documents from >10 vendors and found differing frequencies in the use of doseQuantity, routeCode, administrativeUseCode, indication, and author within Medication Activity. The fact that data elements had the same conformance verb does not mean that they were be included in the CDA document. One may argue that nulling is precisely available for that purpose; but I would push back and say that excessive nulling should be seen as a symptom of a restrictive/inflexible spec or a spec that is progressing faster than its users can keep up.
4. It helps us capture particularly useful pieces of data—For example, the C-CDA IG states that where Guardians are recorded their telecom address may be included. One C-CDA ballot comment suggests changing that to a SHALL conformance because “If the guardian needs to be communicated, having telecom is faster than sending mail to a home address.” As much as having telephone contact details are useful, this is a situation in which the reason for conformance verbs is confused with the convenience perceived by a particular type of implementer.