This is a follow up to a brief discussion on FHIR Chat: https://chat.fhir.org/#narrow/stream/179255-questionnaire/topic/enableWhen.20for.20hiding.20items.3F

In brief: we need a way to conditionally toggle the "readOnly" or "enabled" status of a question based on responses to other question(s).

Description of my use case (from the FHIR Chat):

Let's say I'm building a Questionnaire for patients to fill out when they arrive at a hospital.

This Questionnaire would include questions about their background. Let's say one of those question is, "Do you have a legal guardian?". If yes, we want to ask questions about that guardian, like "First Name," "Last Name," etc.

If the answer is not "yes", then I don't want to clutter the UI by rendering irrelevant "Guardian First Name"-like questions.

This Questionnaire would also include questions about their medical history. There might be a whole section that is a long list of medical conditions, of which an individual is likely to have at most one or two.

Because it is a long, mostly irrelevant list, I want to include a "No to all" option for UX purposes. If this option is selected, I want to lock all of the listed condition responses into being "No" (effectively making them readOnly, or disabled). I'd rather not hide them altogether, because I want to make it obvious that the patient is submitting the fact that they do not have any one of these conditions. And the inverse is also true: I'd like to disable, but not hide, the "No to all" option if any options are selected as "Yes".

Here's what that a Questionnaire that comes close to this functionality would look like:

{
  "resourceType": "Questionnaire",
  "item": [
    {
      "linkId": "patient-background",
      "type": "group",
      "text": "Background",
      "item": [
        {
          "linkId": "has-guardian",
          "text": "Do you have a legal guardian?",
          "type": "boolean"
        },
        {
          "linkId": "guardian-first-name",
          "text": "Guardian First Name",
          "type": "string",
          "enableWhen": [
            {
              "question": "has-guardian",
              "operator": "=",
              "answerBoolean": true
            }
          ]
        },
        {
          "linkId": "guardian-last-name",
          "text": "Guardian Last Name",
          "type": "string",
          "enableWhen": [
            {
              "question": "has-guardian",
              "operator": "=",
              "answerBoolean": true
            }
          ]
        }
      ]
    },
    {
      "linkId": "patient-conditions",
      "type": "group",
      "text": "Past Conditions",
      "item": [
        {
          "linkId": "no-past-conditions",
          "text": "No to all",
          "type": "boolean"
        },
        {
          "linkId": "condition-1",
          "text": "<condition 1 description>",
          "type": "boolean",
          "enableWhen": [
            {
              "question": "no-past-conditions",
              "operator": "=",
              "answerBoolean": false
            }
          ]
        },
        {
          "linkId": "condition-2",
          "text": "<condition 2 description>",
          "type": "boolean",
          "enableWhen": [
            {
              "question": "no-past-conditions",
              "operator": "=",
              "answerBoolean": false
            }
          ]
        },
        {
          "linkId": "condition-3",
          "text": "<condition 3 description>",
          "type": "boolean",
          "enableWhen": [
            {
              "question": "no-past-conditions",
              "operator": "=",
              "answerBoolean": false
            }
          ]
        },
        ... more conditions
      ]
    }
  ]
}

As Lloyd McKenzie pointed out, that "disable" functionality is a little different from the intended use of `enableWhen`, so maybe this should be logically structured differently.

Another line of thinking: maybe we could use some extension that augments calculatedExpression so that a "successful calculation" flips the question's status to "readOnly"?

With that extension, each of the "conditions" questions above could have a `calculatedExpression` that evaluates to false if the "No to all" question is true; otherwise, it would not perform any calculation. This new extension would suggest that, when this `calculatedExpression` is successful, the question goes from "readOnly: false" "readOnly: true". Seems a little convoluted; anyone have more elegant ideas here?