The Delegation Log as an Oversight Instrument
Part of Principal Investigator (PI) Training: Delegation and Supervision.
About 4 minutes. Free, in your language, no account needed.
Why this matters
At most sites the delegation log is treated as paperwork: a grid filled in at initiation, signed once, and remembered only when a monitor asks for it. Treated that way it generates findings and prevents nothing.
Treated properly it is the single most useful management document you have — the only place that answers, on one page, who is allowed to do what at my site, and how do I know they can.
What the guideline actually requires
Section 2.3 asks that delegation be documented in a manner proportionate to the trial. It does not prescribe a form, a layout or a signature grid. E6(R3) also introduces an important relaxation: tasks that fall within the normal course of clinical practice, performed by appropriately qualified staff, do not all need to be individually delegated in the same way as trial-specific procedures.
That relaxation is genuinely useful and it is frequently over-read. A hospital phlebotomist drawing blood in the usual way is doing clinical practice. The same phlebotomist processing a pharmacokinetic sample on a protocol-specified timed schedule, into sponsor-supplied tubes, is doing a trial-specific procedure. Draw the line at whether the protocol changes what the person does.
What a working log contains
For each person, four columns and one honest date range.
| Column | The question it answers | The mistake it prevents |
|---|---|---|
| Name, role, signature and initials | Who is this in the source records? | Unattributable entries |
| Specific tasks delegated | What exactly are they authorized to do? | "All study procedures" |
| Start date | From when? | Work performed before delegation |
| End date | Until when? | Staff who left months ago still listed |
| Investigator signature and date | Who authorized it? | Self-delegation |
The column that matters most is specific tasks. A log that says "all study procedures" for six people tells an inspector nothing and tells you nothing. A log that distinguishes conducts the informed consent discussion, performs protocol-specified assessments, enters source data, dispenses investigational product, and assesses adverse events is a control. It lets you look at a row and ask whether that person's training file supports that row.
Using it as an instrument, not a record
Three habits turn the log from an artifact into oversight.
Update it the day the change happens, not at the next monitoring visit. A new fellow starts on Monday; they are added Monday, before they touch a participant. The most common delegation finding in the world is work performed on a date before the delegation date — and it is nearly always caused by back-filling.
Reconcile it against the training file quarterly. Every task on every row should be traceable to evidence the person was trained for it. Fifteen minutes, four times a year.
Reconcile it against the source records. Take five recent visits, list every set of initials that appears, and check each one is on the log for that task on that date. This is exactly the exercise an inspector performs. Performing it yourself first is the difference between finding your own problem and being shown it.
The end date nobody fills in
Sites are diligent about adding people and negligent about removing them. A coordinator who left in March, still listed as delegated in November, is a control that has stopped working — and it may mean their system access is still live, which is a §2.12.10 problem as well as a delegation one.
Make departure a two-line checklist: end-date the log row, revoke the system access. Attach it to whatever your institution already does when someone leaves.
Key takeaways
- §2.3 requires delegation to be documented proportionately; the form is yours to design.
- E6(R3)'s clinical-practice allowance is real but narrow — the test is whether the protocol changes what the person does.
- Delegate specific tasks, never "all study procedures"; the specificity is what makes the log usable.
- Update on the day of the change, reconcile against training files quarterly and against source records periodically, and always close the end date when someone leaves.