The test decks for QRDA 3 contain duplicate patients. The expected aggregates expect these duplicate patients to be present, but the Race/Ethnicity/etc breakdowns do not.

For example, CMS108v5, the test deck contains 60 patients, two of which are duplicated. The expected aggregate results are IPP 60, and DEMOM 60 (other populations omitted for brevity). If I upload this zip file to our system as is, and import the duplicates as separate patients, we calculate the expected aggregate results for each population as expected.

However, the DENOM race counts for code 2131-1 are being reported as a failure by cypress. We submit 9 members, but Cypress expects 8. After tracing the problem down in our system, I found the extra patient to be one of the duplicate patients.

Further, summing the expected PAYER counts yields 58, not 60. Which is correct?

I see this has been address in an older ticket, CYPRESS-758, but I'd like an answer to why Cypress is calculating the results incorrectly. Are we supposed to de-dupe the results when we import the test deck to our software (thus throwing off the population counts from what Cypress expects) or are we supposed to de-dupe when we doing supplemental data aggregation?