CLIN-11 · HIPAA boundary for students

Design the project so that zero protected health information ever leaves the classroom — and write it as a protocol the rest of your cohort can follow next semester. What gets substituted, what gets synthesized, what stays hypothetical, and what simply does not get done?

Your own or a classmate's planned capstone as the subject, plus your division's data policy and any placement agreement your program already operates under. HHS de-identification guidance supplies the vocabulary — the eighteen Safe Harbor identifiers and what Expert Determination means (https://www.hhs.gov/hipaa/for-professionals/special-topics/de-identification/index.html). This brief produces a document; it touches no clinical data at any point, including as an example.

Restating HIPAA instead of designing a workflow. Assuming de-identification is a student-performable step — it is not, and the protocol should say so. Forgetting the informal channels where PHI actually escapes: phone photos, group chats, personal cloud drives, a shoulder-surfed screen at a showcase, a case detail recognizable in a small town. Writing a protocol with no "this project cannot be done" branch.

A cohort protocol: a decision tree from research question to permissible data source; the substitution table (what a student wanted → what they may use instead: published metrics, open dataset, synthetic records, or hypothetical case) with the analytic cost of each substitution stated honestly; the synthetic-data rules, including that synthetic records are labeled synthetic on every artifact; the leak-channel checklist covering devices, chat, storage, and showcase display; and the stop rule — the conditions under which the project is redesigned rather than continued.

CT: mapping a boundary rather than reciting a rule. CR: designing substitutions that preserve the learning. CO: a peer attempts to route around the protocol and the gaps get closed. CM: written for next year's student, not for the teacher grading it. CZ: the patient whose record was never at risk because this document existed.

A compliance officer, health-information manager, or clinical placement coordinator reviews the protocol against the division's real agreements. Standing instruction: reject any step that assumes a student can de-identify records. School-supervised.

R1: one cohort, one pathway, one protocol. R2: another student could follow it unaided. R3: comparator is the division's current practice, with gaps named. R4: informal leak channels covered explicitly. R5: readable by a first-week student. R6: the stop rule exists and has teeth.

(a) A HIPAA summary with a school logo. (b) A protocol that quietly assumes a placement will hand over "just de-identified" data.

Lane A immediately, and it is the brief most worth assigning first in a cohort that will do the rest of this bank. Lane B relevant where Internship placements exist. No verified credit claim.