Data management plans, database lock readiness and a vendor audit

The documents and reviews that decide whether a dataset survives inspection, and a vendor relationship recovered rather than lost.

The data management plan is not paperwork

A data management plan is often written late, to satisfy a requirement, and then never opened again. Written that way it is worth nothing, and the study runs on whatever the data manager happens to remember.

Written properly it is the operating manual for the study. Who does what, in what order, against what timelines. How queries are raised, routed, escalated and closed. What the validation rules are and which are blocking. What the lock criteria are, stated in advance so lock is a test the study either passes or does not, rather than a negotiation at the end.

I have authored these for interventional studies and the test I hold them to is simple: if the data manager left tomorrow, could someone else run the study from this document. If not, it is not finished.

Pre lock review

Database lock is the only irreversible event in data management, and the most common way it goes wrong is that readiness is assessed in the week it is due.

A pre lock review done properly starts long before. Outstanding queries by age and category. Unverified records. Missing pages against the expected visit schedule. Unresolved consistency failures. Unsigned forms. Coding completeness. Each with an owner and a date, tracked down to zero.

The point is not the checklist, every organisation has one. The point is running it early enough that the findings are still fixable, because a query raised six months after a visit frequently cannot be answered at all.

I led this review for an interventional study and closed it with no critical findings outstanding at lock.

The vendor audit nobody wanted

A vendor audit relationship had broken down, with the engagement effectively stalled and both sides entrenched.

The useful observation was that neither party actually wanted the outcome they were heading toward. The vendor did not want to lose the relationship and we did not want to restart a qualification process from zero with someone new.

Reopening it meant separating the substantive findings from the accumulated friction, agreeing what was genuinely non negotiable on our side, and giving the vendor a defined path back rather than a list of grievances. The engagement was recovered.

Most of what is called vendor management is this. Not enforcement, and not leniency, but being specific enough about what you need that the other side can actually deliver it.

The thread

All three of these are the same job. Decide the standard in advance, make the state of compliance visible continuously, and deal with problems while they are still small enough to be dealt with. Quality is almost never recovered at the end. It is either built in from the start or it is a finding.