Inspection Readiness as a Continuous State

Part of Clinical Trial Assistant (CTA) Training: Inspection Readiness.

About 4 minutes. Free, in your language, no account needed.

Why this matters

Inspection readiness is not preparation. It is the condition a properly run study is in all the time, and the only honest test of it is whether a short-notice inspection would be uneventful.

If the answer is no, the gap is not in your preparation. It is in the conduct.

What an inspector is actually establishing

Three things, roughly.

Were participants protected? Consent obtained properly and documented, ethics approval in place before the activity it covered, safety information acted upon, and eligibility respected.

Is the data credible? Source records exist, data is traceable to them, changes have an audit trail, and the investigational product is accounted for.

Was the trial conducted as approved? The protocol in force was followed, deviations were identified and handled, amendments propagated, and delegated tasks were performed by qualified delegated people.

Every document you file is evidence for one of those questions. Knowing which makes filing a judgment rather than a chore.

The continuous version

A study that is continuously ready has a small number of habits.

Documents are filed close to when they are created, so the file's own dates do not tell a story of neglect. Completeness is checked on a schedule rather than before an event. Reconciliations are run and their outputs filed. Deviations are logged as they occur, with the decision recorded. Actions from monitoring visits are closed and the closure evidenced.

None is difficult. All are the sort of thing that slips when a study is busy, which is exactly when the record matters most.

The fourteen-day drill

If notice arrives, the value of a drill is that it tells you what you did not know.

Run it against a date rather than a folder. Pick a participant and a date, and reconstruct from the file alone: which protocol version applied, which consent version they signed and whether it was the approved one, who was delegated to do what that day, which product lot they received and where it came from, and what happened to any deviation recorded.

If you can do that for three participants at two sites without asking anyone, the study is ready. Whatever you could not answer is the list.

The thing not to do

Do not reconstruct the file. The temptation under notice is to fill gaps quickly, and it produces documents created in a fortnight that purport to describe events from a year earlier. Metadata will show it, and a documentation gap will have become a credibility problem.

A gap that is acknowledged and explained is survivable. A gap that somebody papered over is not.

A worked drill

Pick participant 1047 at site 103, screened 12 February, and reconstruct from the file alone.

Which protocol version applied on 12 February at site 103? The site's ethics approval for amendment 2 is dated 30 January and their implementation record says 5 February, so v3.0. Took ninety seconds.

Which consent version did they sign, and was it approved? Consent dated 12 February, version 3.0, Spanish. The ethics approval names v3.0 Spanish. Clean.

Who consented them and were they delegated? Signed by the coordinator. The delegation log lists her for consent from 14 November. Clean.

Which product lot, from where, at what temperature? Dispensing record gives lot B-2291. The accountability log traces it to shipment 4471, received 2 February, logger within range, receipt signed. Clean.

Any deviations? One: a visit four days outside window in April, logged, assessed as non-serious, closed.

Five questions, about six minutes, no colleague consulted. That site is ready.

What a failed drill looks like

The same exercise at site 107 stalls on question one, because the implementation date is not recorded anywhere and the only evidence is an email saying training happened "last week".

That is the finding, and it is worth far more found now than in an inspection. It is also cheap to fix today and impossible to fix under notice, because reconstructing an implementation date eight months later means asking people to remember, and their recollection is not evidence.

Run the drill on three participants at two sites, quarterly. It takes an afternoon and it is the single most informative thing anyone does to a trial file.

Key takeaways