Attached are four copies of the same test patient for CMS 165. One has valid start and end dates for their hypertension diagnosis, one has only a valid start date and has nullFlavor=UNK for the end date, one has nullFlavor=UNK for the start date but a valid end date, and the fourth has nullFlavor=UNK for both dates. The dates that are present line up to overlap the measurement period as described by the measure.

When we run these patients in Cypress, the first three meet the denominator, but the fourth (both dates null) does not. We are trying to trace back through QRDA and CQL IGs as well as QDM and other measure guidance to find supporting documentation to explain this behavior but are having some trouble. Much of the guidance seems to indicate that a null bound on a interval should be interpreted as either the min or max value for that data point, but that would imply that the prevalence period in this case would be essentially infinite and thus overlap the measurement period as expected, placing the patient into the denominator – which is not what's happening in Cypress.

Is there documentation somewhere that states how an interval with null for both low and high bounds should be evaluated in terms of quality measures? Or if that's too generic, is there any that specifies how a diagnosis prevalence period with null low and high should be evaluated?