Source Data Requirements and Data Integrity Attributes

Part of Good Clinical Practice (GCP) — ICH E6(R3): Data Collection, Source Documentation and CRFs.

About 8 minutes. Free, in your language, no account needed.

Why this matters

An inspector cannot watch your visit. They can only read your records. If the record is missing, unreadable or written a week later, the assessment cannot be trusted. If data is not in the source, it did not happen in regulatory terms. Weak source records are one of the most common reasons a site's data are questioned.

What counts as a source record

The International Council for Harmonisation (ICH) Good Clinical Practice (GCP) guideline E6(R3) uses the term "source records". It no longer uses the older term "source document".

The E6(R3) Glossary defines Source Records as:

"Original documents or data (which includes relevant metadata) or certified copies of the original documents or data, irrespective of the media used."

The Glossary then gives examples. These include participants' medical records, notes and charts; data entered by participants themselves, such as electronic patient-reported outcomes (ePROs); records held by pharmacies, laboratories and other facilities; and data from automated instruments such as wearables and sensors.

A Certified Copy is defined as a copy of the original record that has been verified to hold the same information as the original. Verification is by a dated signature or by generation through a validated process. Relevant metadata must be included where applicable. A scan that nobody has verified is not a certified copy.

You must define your source before the trial starts

This is a requirement many sites miss. E6(R3) §2.12.2 states:

"The investigator should define what is considered to be a source record(s), the methods of data capture and their location prior to starting the trial and should update this definition when needed."

In practice this means a written source data location list, agreed before the first participant is screened. For each protocol data point, you record where the source lives and how it is captured. The same section adds that unnecessary transcription steps between the source record and the data acquisition tool should be avoided. Every extra copy is an extra chance to introduce an error.

The attributes E6(R3) actually requires

E6(R3) states the duty before it states the detail. Under §2.12.1, in generating, recording and reporting trial data the investigator should ensure the integrity of the data under their responsibility, "irrespective of the media used". Paper, electronic and device-captured records are held to one standard.

E6(R3) §2.12.2 then states the requirement in words:

"Source records should be attributable, legible, contemporaneous, original, accurate and complete. Changes to source records should be traceable, should not obscure the original entry and should be explained if necessary (via an audit trail)."

The E6(R3) Glossary entry for Data Integrity lists eight attributes:

The Glossary adds the purpose: data integrity means these criteria are met "such that data are fit for purpose". Fitness for purpose is the standard used throughout E6(R3).

ALCOA and ALCOA++ — useful shorthand, not ICH wording

Many organizations teach these attributes as the acronym ALCOA, or as ALCOA+ or ALCOA++.

Be precise about where that acronym comes from. The acronym does not appear anywhere in ICH E6(R3). Never write "ALCOA++ as required by ICH E6(R3)". The guideline states the attributes directly, in words.

The shorthand is still useful, and it maps closely onto the R3 list. ALCOA covers attributable, legible, contemporaneous, original and accurate. The "+" additions cover the rest. But the industry does not apply the "++" consistently. Different companies add complete, consistent, enduring, available, traceable or secure, in different combinations. That inconsistency is itself a reason to learn the R3 list rather than the acronym.

Teach it this way: the eight R3 attributes are the requirement. ALCOA++ is our memory aid.

Corrections and traceability

A correction must never hide what was written first. The rule for paper records has not changed:

In computerized systems the same rule is enforced by the audit trail. E6(R3) defines an Audit Trail as metadata records that show activities, the initial entry and changes to data fields, "by whom, when and, where applicable, why". Do not work around the system to avoid a query.

Confidentiality, coding and protecting the record

Integrity is one half of the duty. Protecting the record, and the person behind it, is the other.

Privacy. §2.12.7 requires the investigator or institution to implement appropriate measures to protect the privacy and confidentiality of participants' personal information, in accordance with the applicable regulatory requirements on personal data protection. E6(R3) names no data protection law, because the law that applies depends on where the trial runs. Your national or regional legislation supplies the detail.

The participant code. §2.12.8 requires data reported to the sponsor to be identified by an unambiguous participant code that the investigator or institution can trace back to the participant. The sponsor works with the code. Only the site can resolve it to a person, which is why the participant identification record stays at the site.

Systems the site runs itself. §2.12.9 covers systems deployed by the investigator or institution to maintain and retain trial data. Those data must be protected from unauthorized access, disclosure, dissemination or alteration, and from inappropriate destruction or accidental loss. A shared login, an unlocked screen in a clinic room, or a spreadsheet of visit data on a personal laptop each fails that test.

In practice

Amara is a coordinator at a site running a Phase III hypertension trial for ABC Pharmaceuticals. Blood pressure is a protocol endpoint. The clinic uses an automated monitor that prints a strip.

Before the first participant, Amara and the investigator write a source data location list. It says the printed strip is the source for blood pressure, it is taped into the visit worksheet, and the value is then entered into the electronic system. The reading is not written on a sticky note first.

At visit 3 the device prints 148/92. Amara writes 148/92 into the worksheet but transposes it as 184/92. She notices at the end of the day. She draws one line through 184, writes 148 beside it, initials, dates, and writes "transcription error, see printed strip". The strip is still attached. Nothing is hidden and the correction is explained.

Compare that with a second site, where a coordinator writes readings on scrap paper during the clinic and copies them into the chart the following week. There is no strip. There is no source data location list. The monitor asks a simple question: what is the source? Nobody can answer it.

Illegible notes can be treated as if the data do not exist. So can undefined ones.

Key takeaways