Finding the National Instrument, Not the Summary

Part of Regulatory Affairs and the Jurisdiction Map: Reading a Jurisdiction.

About 6 minutes. Free, in your language, no account needed.

Why this matters

There is a quick way to answer a regulatory question and a defensible way, and they feel identical while you are doing them.

The quick way finds a consultancy page, a conference slide, a vendor's country guide, or an internal document from a previous study, and reports what it says. The defensible way finds the law, the regulation or the authority's own guidance, and reports what that says, with a citation someone else can check.

The difference shows up at exactly one moment: when an inspector, a partner, or a large sponsor's due diligence team asks where a claim came from. At that moment, a secondary source is not an answer.

What counts as a primary source

In descending order of authority:

The legal instrument itself. An act, a regulation, a statutory rule, a ministerial order. In CRLN's verified register these are named precisely, and the precision is the point: not "Indian clinical trial rules" but the New Drugs and Clinical Trials Rules, 2019, First Schedule. Not "South African GCP" but the Medicines and Related Substances Act 101 of 1965, Regulation 30, with the South African Good Clinical Practice guidelines, third edition 2020.

The authority's own published guidance. Issued by the regulator, on the regulator's own publication channel. This is genuinely primary even where it is formally advisory, and in some cases it is explicitly compulsory.

Official multilateral records. The ICH membership register for participation status. The World Health Organization's published benchmarking outcomes for maturity level. These are primary for the narrow facts they record and nothing else.

Everything else is secondary: law firm briefings, industry association summaries, CRO country guides, academic reviews, previous study documents, and our own notes. All of these are useful for orientation. None of them is the basis for a decision.

Why the summary is usually wrong in a specific way

Secondary sources do not fail randomly. They fail in a pattern.

They are behind. A country guide written in 2023 describes the framework as it was in 2023. If the instrument was amended since, the guide is confidently wrong and carries no signal that it is.

They generalize the region. A summary covering twelve countries in a region will describe the common pattern and quietly omit the two countries that deviate, which are the two you needed.

They omit the conditions. A rule is reported without the exception that applies to your trial, because the exception was not relevant to whoever wrote the summary.

They lose the scope. A benchmarking level is reported as a number without the product scope it was assessed for, which changes what it means entirely.

Each of these produces an answer that is right enough to pass a casual review and wrong enough to matter.

How to verify efficiently

You are not reading a legal code. You are answering specific questions, and the discipline is to know which ones.

Start from the authority's own site and find its clinical trials section. Identify the instrument it cites as the basis for its authority, because the authority will tell you. Record the instrument's name, its number, its year, and the specific part that applies. Note the date you checked, because that date is what makes the record auditable later.

Then work the seven categories where national law reliably adds to the baseline: consent by representative, emergency consent, safety clocks and recipients, import and handling of product, data protection, insurance and compensation, and post-trial obligations.

If a question cannot be answered from the authority's own published material, that is itself a finding. Record it as unverified rather than filling the gap with an inference from a neighbor.

In practice

A regulatory associate is asked whether a particular country requires a local sponsor representative.

The quick path finds a CRO country guide saying yes, and reports yes. Total time, eleven minutes.

The defensible path goes to the authority's site, finds the clinical trial application guidance, and discovers that the requirement exists but applies only where the sponsor has no legal presence in a defined economic community, and that the sponsor in question does have one through a subsidiary. The answer is no, with a condition, and a citation. Total time, ninety minutes.

The eleven minute answer would have triggered three months of work establishing a representative that was not required. The ninety minute answer saved that, and it is the answer that survives the question "where did you get that?"

Note what makes the difference: not diligence in the abstract, but going to the instrument, where conditions live. Conditions are exactly what summaries strip out.

What this looks like when it goes wrong

A sponsor's global regulatory intelligence file describes a country's requirements. It is detailed, well formatted, and used by every study the sponsor runs there.

It was compiled in 2021 from a mixture of a law firm briefing and an earlier internal file, and nobody recorded which claims came from which source or when each was last checked. It has been updated twice, by different people, each of whom edited some entries and not others.

There is now no way to tell, for any given line, whether it reflects current law, 2021 law, or an inference someone made. The file cannot be corrected without redoing all of it, and it cannot be trusted without correcting it. Every study that used it inherited the uncertainty.

The failure was not using a secondary source. It was not recording provenance and date, which makes a source impossible to re-verify and therefore impossible to age out.

Key takeaways