When Your Research Mentor Moves to Another Institution Mid-Project: How to Salvage Your Data

When your research mentor drops the "I'm moving to Stanford in 6 weeks" bomb in the middle of your chart review, your first reaction is panic. Second reaction should be triage. Let me break this down specifically: you do not have 6 weeks. You have about 10 business days before administrative access locks start to shift and your project becomes institutional driftwood.

I have seen this happen three times to students I advise. One lost 14 months of work because she assumed a verbal "yeah, keep working on it" over Zoom counted as permission. It does not. The system is built to protect institutions, not your poster presentation.

This situation is salvageable. But only if you stop thinking like a mentee and start thinking like a compliance officer.

This article is for educational purposes only and does not constitute legal advice. It is not financial advice and it is not tax advice. Institutional policies and figures vary, so consult a qualified professional at your home institution before transferring any data.

The Mid-Project Exodus: Assessing the Immediate Landscape

The departure shock is real. You have spent months building a REDCap database, pulling charts, running your first regressions. Suddenly your PI is packing their lab. Your job now is not to mourn. It is to audit.

Three legal realities you must separate immediately:

1. IRB ownership. The IRB protocol belongs to the institution, not the PI. Your home institution's IRB approved the study. The PI is listed as the individual of record, but the assurance is between the hospital/university and the federal government. When the PI leaves, the protocol does not automatically travel with them. It either closes, transfers to a new local PI, or requires a formal reliance agreement.

2. Data governance. This is where most students get burned. That Excel sheet with 300 patients, MRNs, and outcomes? That is Protected Health Information housed on an institutional server. You do not own it. Your PI does not own it either in a way that allows free transport. The institution does. Taking it on a USB drive, emailing it to your mentor's Gmail, dropping it in a personal Dropbox, all violations. Potentially reportable violations.

3. Intellectual property. The idea, the preliminary data, the abstract you submitted? That belongs to the institution under most faculty appointment contracts. The new institution will claim any future work product generated there. This creates a gray zone for your half-finished manuscript.

So, direct assessment. Open three tabs right now: your IRB protocol in the IRB system, your data storage location, and your mentor's appointment end date. Document everything. Screenshot your current IRB approval letter, personnel list, and data security plan.

Now, feasibility. Remote mentorship versus local transition. Let me be blunt: remote mentorship almost always fails unless you secure local anchoring. Why? You cannot order charts. You cannot access the EMR from the outside without affiliation. You cannot submit modifications. You need a local faculty member with an MD or PhD after their name in your own department who is willing to become PI of record.

Remote mentorship works for writing, statistics, editing. It does not work for data access and regulatory compliance. You need both models running in parallel. Your departing mentor as senior author/collaborator. A new local co-investigator as the IRB principal.

Do not wait until your mentor is gone to find this person. The department chair is your best ally here, not your enemy.

Navigating Institutional Review Board (IRB) and Data Use Agreements

This is the bottleneck that kills 80% of these transitions. Let me break this down specifically: IRB staff move slowly. DUA offices move even slower. If your mentor leaves on June 30th, you need to have your amendment submitted by mid-June at the latest.

Here is the exact workflow you should follow:

Step 1: Audit the protocol. Is it expedited? Is it continuing review exempt? Who else is listed as key personnel? Are you listed? If you are not listed as key personnel, you have zero standing. Fix that first with an amendment adding you if possible.

Step 2: Decide on data disposition. Two paths.

Path A: Data stays local. This is cleaner. You keep your REDCap project, your analytic files, your chart access under a new local PI. Your departing mentor receives only de-identified, limited datasets under a DUA for remote collaboration. Most institutions prefer this.

Path B: Data moves. Your mentor wants to take the dataset to continue enrollment at their new site. This requires a formal Data Use Agreement, sometimes a Material Transfer Agreement if biospecimens are involved, and often an IRB reliance agreement or a new IRB protocol at the receiving institution. Expect 6-12 weeks minimum. I have seen DUAs take 4 months because university legal offices argue over indemnification language.

Step 3: Chain of custody for PHI. You must never hold PHI on a personal device. Your technical salvage plan must document: where was PHI originally stored? Encrypted institutional drive? REDCap? Where will de-identified data be stored now? How will you de-identify, Safe Harbor or expert determination? Who has the linking file? Who destroys it and when?

Write this down in a one-page data security memo. IRB coordinators love clarity. Vagueness gets your amendment returned.

Pro tip: If your dataset already contains only de-identified data and you have documentation that it was de-identified appropriately, the transfer is dramatically easier. If you are still in chart abstraction pulling MRNs, you are in a much weaker position. Push to finish extraction and de-identify before the exit date.

Communication Protocols: Engaging the Departed Mentor vs. The Department Chair

This is politically delicate. Your mentor feels guilty but distracted. The chair feels annoyed and suddenly responsible. You are in the middle.

You need two different emails. Not one group email. Two tailored communications.

Email 1: To your departing mentor. Short. Professional. Focused on preservation.

You are asking for three things explicitly:

1. Written confirmation of your role and continued authorship (e.g., "I would like to confirm I will remain first author on manuscript X based on my contribution to data collection and analysis").
2. Willingness to sign a letter of collaboration or continue as formal co-mentor / senior author after the move.
3. Introduction to a local faculty member they trust to take over as local PI.

Bad example: "Can I still publish with you?" That is vague and easy to ignore.

Good example: "Dr. [Name], to ensure continuity after your transition on [Date], could you confirm: (a) you support me continuing as first author on the [Topic] manuscript, (b) you are willing to remain as senior author/collaborator via remote mentorship, and (c) you would support Dr. [Local Faculty] as local PI of record for IRB continuity?"

Get this in writing. Email counts as writing. If they ghost you, that is data. Escalate to the chair.

Email 2: To the department chair or division chief. You are not complaining. You are requesting institutional continuity.

Frame it as risk mitigation for the department, not just your CV. Chairs do not want an open IRB protocol with no PI. That is an audit vulnerability. You are offering a solution.

"Dear Dr. [Chair], as you may be aware Dr. [Mentor] is transitioning effective [Date]. I am currently listed as key personnel on IRB #XXXXX, with approximately 80% data collection complete. To prevent a lapse in protocol oversight and preserve the department's data asset, I am requesting assignment of a local faculty sponsor to serve as PI of record. Dr. [Suggested Name] is familiar with the project. I have drafted an IRB personnel amendment for review."

This is not presumptuous. This is competent. Chairs appreciate trainees who solve administrative problems.

Now, authorship protection. This is where projects die quietly. Your mentor moves, new fellows at the new institution redo the analysis, your name slides to third. Or the new local PI at your institution decides to add their own resident and take first authorship.

Prevention is a written authorship agreement early. ICMJE criteria are clear. Document who did what. If you collected the data, built the database, did the primary analysis, you meet first author criteria. Send a polite authorship proposal table: Author 1, you, data curation, formal analysis, writing original draft; Author Last, departing mentor, conceptualization, supervision; etc.

If you meet resistance, loop in the research integrity officer or ombudsman. Not as a threat. As a clarification.

Data Salvage Operations: Technical and Statistical Continuity

Let me break this down specifically: the technical loss is more predictable than the political loss. And it is entirely preventable if you act before access revocation.

Rule 1: Compliant backup. Do not download PHI to your laptop. Ever. Compliant backup means ensure the data lives in at least two institutional-approved locations with versioning. REDCap + institutional Box/Google Drive with encryption. Or a lab server with nightly backup. Confirm with your IRB's data security plan what is allowed. If your data is on your mentor's personal C: drive with no backup, that is the first emergency to fix.

Rule 2: Extract your code pipeline. This is what most students forget. Your data is useless without the code that cleaned it. I have seen students lose everything because the R scripts lived on their PI's desktop computer that was wiped after departure. Ouch.

Today: push every.R,.py,.do,.sps file to a private GitHub repository or OSF project that is approved by your institution. Include a README that says exactly how to run the pipeline from raw export to analytic dataset. Include package versions with renv or conda environment.yml. Clone that repo to the new local PI's server environment.

Rule 3: Prepare for a split team manuscript. Writing across institutions is slower. Establish early: who holds the master Word document? Google Docs is not always HIPAA compliant if you paste table shells with PHI. Use Overleaf or institutional OneDrive with tracked changes. Set a biweekly 30-minute Zoom for manuscript check-ins. No agenda, no meeting.

Delays are predictable. Real data:

Eight weeks for IRB personnel changes. Six weeks for DUA negotiation. Three weeks for code migration. That is 17 weeks of preventable delay if you start late. If you start the day your mentor announces, you can compress that to 3-4 weeks total.

Final action steps. This is your immediate checklist for the next 72 hours:

1. Pull your IRB approval letter, personnel list, and data security plan. Save PDFs.
2. Inventory all data: location, identifiability, backup status, access credentials.
3. Export and securely store all analytic code with documentation in an approved repository.
4. Send the two high-yield emails described above. Get written confirmation of authorship and future roles.
5. Meet with your IRB coordinator in person. Not email. In person. Explain the timeline. Ask what they need for a PI change amendment.
6. File a formal DUA request through your technology transfer office if any data will leave the institution, even de-identified.

Your mentor moving is not the end of your project. Treated like an informal favor, it will become a disaster. Treated like a formal administrative transition with paperwork, agreements, and local anchoring, it becomes just another systems-based practice lesson medicine loves to teach you.

Do the boring administrative work now. Publish later.