This proposal was discussed with a standards group consisting of representatives from MITRE, Lantana, and Telligen (MAT). While there is some discomfort with the inconsistent treatment of missing end dates, the group agreed to move forward with the recommendation of the QDM User Group (the benefits outweighing the concerns). At some point in the future, we may consider narrowing the "fuzziness" of the end date calculation to specific datatypes (such as those with an "active" status).

While the user group did reach consensus on this issue during the user group meeting, we also agreed that we would put forth a final email for confirmation. If we do not hear any dissenting opinions by the end of the week, we will bring this recommendation to the MCCB for final approval. After that, Bonnie will update its execution strategy to match the recommendation of the user group.

This issue was discussed during the August 2014 QDM User Group meeting.

The User Group noted that this concept is used several times throughout Meaningful Use measures. If the original intent of “overlaps” (to replace the previous more complex logic) isn't viable, then the value and use of "overlaps" is significantly limited.

The User Group recommends continuing with overlaps supporting missing end dates. Missing end dates would be interpreted as an event that is ongoing. However, guidance or warning of this calculation should be clearly stated since it is inconsistent with other operators (e.g., during). This approach matches the original intent of overlaps. This recommendation, however, should be confirmed by the standards group that meets on Tuesdays. After that, it will be brought to the MCCB.

In addition, there was some concern about events that have no end date but should not be interpreted as "ongoing" (e.g., "point-in-time" events). It was suggested that in order to reduce unexpected results, vendors should record "point-in-time" events with an effectiveTime low and high, where the low and high are the same. This allows for a distinction between "point-in-time" events and events that are "ongoing".

Confirming that QDM-1 does seem to indicate that the left-hand side (LHS) can have a null end date:

Additionally, if either X or Y had null start or end dates, the evaluation of X overlaps Y would evaluate equivalently with the original logical statements above. In other words, as long as X starts before the end of Y and X does not end before the start of Y (including a null end date on X), then overlaps would evaluate to true.

The Bonnie team has confirmed that if the LHS has a null end date, they evaluate it as false – which is not consistent with QDM-1 or the previous SBS AND NOT EBS construct. The Bonnie team implemented it like this based on the description in the QDM 4.1 specification:

Since events can have a left overlap, right overlap, inner overlap, or outer overlap, the Overlaps
operator uses the start date/time and end/time of each event to determine if there is overlap.

The Bonnie team has indicated that they can special case overlaps to evaluate to true when the LHS end date is null, but wouldn't recommend expanding this out to other operators (like during) since it would affect existing logic. The QDM User Group needs to determine if they want Bonnie to make overlaps work the same as the previous SBS AND NOT EBS construct.

Part of the confusion arises from the ambiguity about what a null end date means. Should it be interpreted as "end date not known" or "end date is ongoing"? Some QRDA templates seem to indicate that lack of an end date implies it is ongoing (e.g., in a diagnosis). HQMF seems to suggest that ongoing should be represented by the PINF nullFlavor – and also suggests that if an end date is truly null, that the HQMF overlap operator should evaluate to false.

We will bring this to the QDM User Group to determine how the community wishes end dates to be interpreted in the context of the overlaps temporal operator.