I was showing off Atlas demo when an interesting question was asked. How can the DPC be > 1000 when utilizing SYNPUF1K? Is SYNPUF1K not 1000 person dataset?
This is a limitation of how the aggregate rollups work: if a person appears in 2 different descendant child groups (such as below ingredient there are 2 forms which the person exposed to both) we will roll up those 2 counts to the parent, double counting. This is just a limitation and tradeoff between simplicity and accuracy. The RC/DRC/PC/DPC shouldn’t be considered an exact count, but rather used as a heuristic about the presence of codes in your data.