Under the Safe Harbor method an extract is de-identified when the 18 identifiers are removed and the organization has no actual knowledge that what remains could identify a person. In SQL that means rewriting columns, not just dropping them:
- ZIP code: keep the first three digits only where the three-digit area has more than 20,000 people; otherwise 000. The populations are in
hd_zip3_population.
- Dates: all elements of dates except the year. A month is an element of a date.
- Ages over 89: grouped as 90 or older, and the birth year of those patients goes too, since it gives their age away.
- Direct identifiers (names, MRN, phone, email and the rest) leave the extract.
CASE WHEN z.population > 20000 THEN z.zip3 ELSE '000' END AS zip3
The Safe Harbor ZIP rule in SQL.