eduKateSG · Why Science?
Let a clinical-trial datum tell you where it came from and what happened next
Read FDA’s September 2013 final eSource guidance through data originators, direct entry, eCRFs, identifiers, investigator review, corrections, retention and inspection.
A clinical-trial result may begin as a nurse’s observation, a participant’s answer, a laboratory instrument reading or a device measurement. Electronic source data can remove duplicate transcription, but only when each value keeps enough context to be attributed and checked.
FDA’s final September 2013 guidance addresses source data used to fill predefined fields in an electronic case report form, or eCRF. It promotes electronic capture while protecting reliability, quality, integrity and traceability from source to regulatory submission.
Did you know? FDA uses “data originator” for the person, system, device or instrument authorised to enter, change or transmit a data element into the eCRF. The person who reviews a value is not automatically the person or system that originated it.
For students, eSource is a practical lesson in research skills: record an observation once, label its origin, keep changes visible and preserve enough evidence for someone else to reconstruct the experiment.
Chapter index
Chapters 1–9: source data, originators, identifiers and direct entry
- 1. Electronic source data begin in electronic form
- 2. The September 2013 guidance is final and cross-centre
- 3. An eCRF is an auditable study record
- 4. The data originator can be a person or a machine
- 5. The protocol should identify authorised originators
- 6. Direct entry should avoid unnecessary duplication
- 7. Data element identifiers carry origin metadata
- 8. Automatic transfer should preserve the real source
- 9. Investigator access supports contemporaneous oversight
Chapters 10–17: review, corrections, retention, inspection and reconstruction
- 10. Investigator review should be documented
- 11. Corrections should retain old and new values
- 12. Queries and sponsor review should not rewrite source
- 13. Computerised systems need controlled descriptions
- 14. Retention should survive system and vendor change
- 15. FDA inspection needs direct, understandable access
- 16. eSource does not erase participant privacy duties
- 17. A complete eSource receipt makes the trial reproducible
FDA defines electronic source data as source data initially recorded in electronic format. Source data include original records and certified copies used to reconstruct and evaluate a clinical investigation. The definition therefore concerns the first trustworthy recording, not merely the point at which a paper value is later typed into a database.
Follow electronic source data begin in electronic form like a laboratory observation travelling through a network. FDA defines electronic source data as source data initially recorded in electronic format. Source data include original records and certified copies used to reconstruct and evaluate a clinical investigation. The definition therefore concerns the first trustworthy recording, not merely the point at which a paper value is later typed into a database. The trustworthy conclusion keeps this limit beside it: Scanning or transcribing a paper record later does not make the original observation electronic source data.
Try this
Build a provenance card for electronic source data begin in electronic form. Choose one trial value and identify the first place where it is recorded. Then state whether that record is original, a certified copy or a later transcription. Add one failure signal and one person responsible for checking it. Keep this limit visible: Scanning or transcribing a paper record later does not make the original observation electronic source data.
A strong science note separates observation from interpretation. Choose one trial value and identify the first place where it is recorded. Then state whether that record is original, a certified copy or a later transcription. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Choose one trial value and identify the first place where it is recorded. Then state whether that record is original, a certified copy or a later transcription. Keep this boundary visible: Scanning or transcribing a paper record later does not make the original observation electronic source data.
CHAPTER 2 OF 17 · Name the source
2. The September 2013 guidance is final and cross-centre
FDA identifies the document as final guidance dated September 2013 under docket FDA-2010-D-0643. It was issued by CDER, CDRH and CBER for sponsors, contract research organisations, investigators and others involved in capture, review and retention of electronic source data.
For the september 2013 guidance is final and cross-centre, provenance is the scientific story. FDA identifies the document as final guidance dated September 2013 under docket FDA-2010-D-0643. It was issued by CDER, CDRH and CBER for sponsors, contract research organisations, investigators and others involved in capture, review and retention of electronic source data. That story remains bounded because Guidance states FDA’s current thinking and does not replace applicable statutes, regulations or newer topic-specific guidance.
Try this
Turn the september 2013 guidance is final and cross-centre into a before-and-after audit. Create a source receipt with title, final status, date, docket, issuing centres, audience and scope. Explain what evidence would reveal an incorrect transfer or correction. Respect this boundary: Guidance states FDA’s current thinking and does not replace applicable statutes, regulations or newer topic-specific guidance.
A strong science note separates observation from interpretation. Create a source receipt with title, final status, date, docket, issuing centres, audience and scope. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Create a source receipt with title, final status, date, docket, issuing centres, audience and scope. Keep this boundary visible: Guidance states FDA’s current thinking and does not replace applicable statutes, regulations or newer topic-specific guidance.
FDA describes an electronic case report form as an auditable electronic record of information reported to the sponsor for each participant according to the protocol. It supports systematic capture, review, management, storage, analysis and reporting. The eCRF is one record in a wider data pathway, not the entire clinical system.
A useful test of an ecrf is an auditable study record is whether a new reviewer can reconstruct it without asking the original team. FDA describes an electronic case report form as an auditable electronic record of information reported to the sponsor for each participant according to the protocol. It supports systematic capture, review, management, storage, analysis and reporting. The eCRF is one record in a wider data pathway, not the entire clinical system. Preserve this caution: An eCRF field can be complete yet still lack adequate evidence about origin, meaning or correction.
Try this
Teach an ecrf is an auditable study record using claim, source, transformation and limitation. Map one eCRF field to its protocol purpose, originator, source record, edit rules and downstream analysis. Let a partner challenge the weak link. Do not lose this guardrail: An eCRF field can be complete yet still lack adequate evidence about origin, meaning or correction.
A strong science note separates observation from interpretation. Map one eCRF field to its protocol purpose, originator, source record, edit rules and downstream analysis. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Map one eCRF field to its protocol purpose, originator, source record, edit rules and downstream analysis. Keep this boundary visible: An eCRF field can be complete yet still lack adequate evidence about origin, meaning or correction.
CHAPTER 4 OF 17 · Name the source
4. The data originator can be a person or a machine
A data originator may be a participant, investigator, delegated staff member, computer system, device or instrument authorised to enter, change or transmit a data element. Identifying originators supports attribution and helps distinguish direct measurements from reviewed, derived or transcribed values.
Follow the data originator can be a person or a machine like a laboratory observation travelling through a network. A data originator may be a participant, investigator, delegated staff member, computer system, device or instrument authorised to enter, change or transmit a data element. Identifying originators supports attribution and helps distinguish direct measurements from reviewed, derived or transcribed values. The trustworthy conclusion keeps this limit beside it: The account receiving a transmitted value is not necessarily the originator of the observation.
Try this
Build a provenance card for the data originator can be a person or a machine. For five sample data elements, name the originator type and the evidence that proves the attribution. Add one failure signal and one person responsible for checking it. Keep this limit visible: The account receiving a transmitted value is not necessarily the originator of the observation.
A strong science note separates observation from interpretation. For five sample data elements, name the originator type and the evidence that proves the attribution. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
For five sample data elements, name the originator type and the evidence that proves the attribution. Keep this boundary visible: The account receiving a transmitted value is not necessarily the originator of the observation.
CHAPTER 5 OF 17 · Capture the datum
5. The protocol should identify authorised originators
FDA recommends that sponsors list or describe authorised data originators in the protocol and may use a separate document referenced by the protocol when details change. The list should be controlled so that roles remain current without silently rewriting who was authorised at an earlier time.
For the protocol should identify authorised originators, provenance is the scientific story. FDA recommends that sponsors list or describe authorised data originators in the protocol and may use a separate document referenced by the protocol when details change. The list should be controlled so that roles remain current without silently rewriting who was authorised at an earlier time. That story remains bounded because A flexible role list must still preserve version history and site-specific accountability.
Try this
Turn the protocol should identify authorised originators into a before-and-after audit. Build an originator matrix with data element, role or system, authorisation date, version and retirement date. Explain what evidence would reveal an incorrect transfer or correction. Respect this boundary: A flexible role list must still preserve version history and site-specific accountability.
A strong science note separates observation from interpretation. Build an originator matrix with data element, role or system, authorisation date, version and retirement date. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Build an originator matrix with data element, role or system, authorisation date, version and retirement date. Keep this boundary visible: A flexible role list must still preserve version history and site-specific accountability.
CHAPTER 6 OF 17 · Capture the datum
6. Direct entry should avoid unnecessary duplication
Direct electronic capture can allow an observation to enter the eCRF without first being recorded elsewhere. FDA’s guidance encourages approaches that reduce duplicate entry and transcription error while keeping the initial record attributable, contemporaneous and available to authorised reviewers.
A useful test of direct entry should avoid unnecessary duplication is whether a new reviewer can reconstruct it without asking the original team. Direct electronic capture can allow an observation to enter the eCRF without first being recorded elsewhere. FDA’s guidance encourages approaches that reduce duplicate entry and transcription error while keeping the initial record attributable, contemporaneous and available to authorised reviewers. Preserve this caution: Fewer copies improve efficiency only when the retained electronic record is adequate as source.
Try this
Teach direct entry should avoid unnecessary duplication using claim, source, transformation and limitation. Compare a paper-to-eCRF route with direct entry and mark every transcription, verification and retained record. Let a partner challenge the weak link. Do not lose this guardrail: Fewer copies improve efficiency only when the retained electronic record is adequate as source.
A strong science note separates observation from interpretation. Compare a paper-to-eCRF route with direct entry and mark every transcription, verification and retained record. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Compare a paper-to-eCRF route with direct entry and mark every transcription, verification and retained record. Keep this boundary visible: Fewer copies improve efficiency only when the retained electronic record is adequate as source.
CHAPTER 7 OF 17 · Capture the datum
7. Data element identifiers carry origin metadata
FDA discusses identifiers that can help show the origin of each data element and facilitate audit-trail examination. Useful metadata may identify who or what generated a value, the date and time, and whether the element was entered directly, imported or derived.
Follow data element identifiers carry origin metadata like a laboratory observation travelling through a network. FDA discusses identifiers that can help show the origin of each data element and facilitate audit-trail examination. Useful metadata may identify who or what generated a value, the date and time, and whether the element was entered directly, imported or derived. The trustworthy conclusion keeps this limit beside it: An identifier without a documented meaning cannot support reconstruction.
Try this
Build a provenance card for data element identifiers carry origin metadata. Design an identifier legend for person entry, device transfer, laboratory import and calculated value. Add one failure signal and one person responsible for checking it. Keep this limit visible: An identifier without a documented meaning cannot support reconstruction.
A strong science note separates observation from interpretation. Design an identifier legend for person entry, device transfer, laboratory import and calculated value. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Design an identifier legend for person entry, device transfer, laboratory import and calculated value. Keep this boundary visible: An identifier without a documented meaning cannot support reconstruction.
CHAPTER 8 OF 17 · Capture the datum
8. Automatic transfer should preserve the real source
When data move from an instrument, device, EHR or other system into the eCRF, the transfer should preserve origin and avoid making an automated measurement appear manually entered. Interfaces, mappings, units and timestamps need control because a technically successful transfer can still alter scientific meaning.
For automatic transfer should preserve the real source, provenance is the scientific story. When data move from an instrument, device, EHR or other system into the eCRF, the transfer should preserve origin and avoid making an automated measurement appear manually entered. Interfaces, mappings, units and timestamps need control because a technically successful transfer can still alter scientific meaning. That story remains bounded because Automation reduces transcription error but can repeat a mapping error consistently at scale.
Try this
Turn automatic transfer should preserve the real source into a before-and-after audit. Trace one automated value through unit conversion, field mapping, transmission, acceptance and display. Explain what evidence would reveal an incorrect transfer or correction. Respect this boundary: Automation reduces transcription error but can repeat a mapping error consistently at scale.
A strong science note separates observation from interpretation. Trace one automated value through unit conversion, field mapping, transmission, acceptance and display. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Trace one automated value through unit conversion, field mapping, transmission, acceptance and display. Keep this boundary visible: Automation reduces transcription error but can repeat a mapping error consistently at scale.
CHAPTER 9 OF 17 · Review the record
9. Investigator access supports contemporaneous oversight
The clinical investigator should have access to electronic source data during and after the investigation as required for review, retention and FDA inspection. Access should allow the investigator to see data relevant to participant safety and protocol conduct rather than receiving an opaque summary after the fact.
A useful test of investigator access supports contemporaneous oversight is whether a new reviewer can reconstruct it without asking the original team. The clinical investigator should have access to electronic source data during and after the investigation as required for review, retention and FDA inspection. Access should allow the investigator to see data relevant to participant safety and protocol conduct rather than receiving an opaque summary after the fact. Preserve this caution: Access to a dashboard does not prove that critical source data were reviewed in time.
Try this
Teach investigator access supports contemporaneous oversight using claim, source, transformation and limitation. List the data the investigator must see, the review frequency, evidence of review and escalation threshold. Let a partner challenge the weak link. Do not lose this guardrail: Access to a dashboard does not prove that critical source data were reviewed in time.
A strong science note separates observation from interpretation. List the data the investigator must see, the review frequency, evidence of review and escalation threshold. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
List the data the investigator must see, the review frequency, evidence of review and escalation threshold. Keep this boundary visible: Access to a dashboard does not prove that critical source data were reviewed in time.
FDA discusses investigator review and electronic signing of the completed eCRF. The review process should show what was reviewed, by whom, when and under which version or state. Review may be risk-based, but critical data and unresolved queries should not disappear behind a blanket certification.
Follow investigator review should be documented like a laboratory observation travelling through a network. FDA discusses investigator review and electronic signing of the completed eCRF. The review process should show what was reviewed, by whom, when and under which version or state. Review may be risk-based, but critical data and unresolved queries should not disappear behind a blanket certification. The trustworthy conclusion keeps this limit beside it: A final signature cannot retroactively repair unsafe delays or unexplained data changes.
Try this
Build a provenance card for investigator review should be documented. Create a review receipt linking the signer, record state, open queries, date and certification meaning. Add one failure signal and one person responsible for checking it. Keep this limit visible: A final signature cannot retroactively repair unsafe delays or unexplained data changes.
A strong science note separates observation from interpretation. Create a review receipt linking the signer, record state, open queries, date and certification meaning. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Create a review receipt linking the signer, record state, open queries, date and certification meaning. Keep this boundary visible: A final signature cannot retroactively repair unsafe delays or unexplained data changes.
CHAPTER 11 OF 17 · Review the record
11. Corrections should retain old and new values
Changes to electronic source data should be attributable and traceable. The record should preserve the original value, corrected value, person making the change, date and time, and a reason when appropriate. Corrections should not obscure the history needed to evaluate the trial.
For corrections should retain old and new values, provenance is the scientific story. Changes to electronic source data should be attributable and traceable. The record should preserve the original value, corrected value, person making the change, date and time, and a reason when appropriate. Corrections should not obscure the history needed to evaluate the trial. That story remains bounded because Deleting a wrong value without its history removes evidence needed to understand how the record evolved.
Try this
Turn corrections should retain old and new values into a before-and-after audit. Reconstruct one correction and ask whether a reviewer can see the trigger, decision and effect on analysis. Explain what evidence would reveal an incorrect transfer or correction. Respect this boundary: Deleting a wrong value without its history removes evidence needed to understand how the record evolved.
A strong science note separates observation from interpretation. Reconstruct one correction and ask whether a reviewer can see the trigger, decision and effect on analysis. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Reconstruct one correction and ask whether a reviewer can see the trigger, decision and effect on analysis. Keep this boundary visible: Deleting a wrong value without its history removes evidence needed to understand how the record evolved.
CHAPTER 12 OF 17 · Review the record
12. Queries and sponsor review should not rewrite source
Sponsors and monitors may identify inconsistencies and ask sites to resolve them. The investigator or authorised site personnel should make source corrections with a visible audit trail. Sponsor access should be designed to prevent unauthorised alteration of site source records.
A useful test of queries and sponsor review should not rewrite source is whether a new reviewer can reconstruct it without asking the original team. Sponsors and monitors may identify inconsistencies and ask sites to resolve them. The investigator or authorised site personnel should make source corrections with a visible audit trail. Sponsor access should be designed to prevent unauthorised alteration of site source records. Preserve this caution: A query can prompt a correction; it should not silently author the site’s clinical observation.
Try this
Teach queries and sponsor review should not rewrite source using claim, source, transformation and limitation. Separate query creation, clinical judgement, source correction, verification and closure into accountable steps. Let a partner challenge the weak link. Do not lose this guardrail: A query can prompt a correction; it should not silently author the site’s clinical observation.
A strong science note separates observation from interpretation. Separate query creation, clinical judgement, source correction, verification and closure into accountable steps. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Separate query creation, clinical judgement, source correction, verification and closure into accountable steps. Keep this boundary visible: A query can prompt a correction; it should not silently author the site’s clinical observation.
CHAPTER 13 OF 17 · Preserve the trail
13. Computerised systems need controlled descriptions
FDA recommends describing computerised systems used to create, modify, maintain, archive, retrieve or transmit source data. Documentation should make system purpose, interfaces, security, access, backup and change control understandable to investigators, sponsors and inspectors.
Follow computerised systems need controlled descriptions like a laboratory observation travelling through a network. FDA recommends describing computerised systems used to create, modify, maintain, archive, retrieve or transmit source data. Documentation should make system purpose, interfaces, security, access, backup and change control understandable to investigators, sponsors and inspectors. The trustworthy conclusion keeps this limit beside it: A vendor brochure does not describe the deployed configuration or the study-specific data flow.
Try this
Build a provenance card for computerised systems need controlled descriptions. Write a system card with name, version, owner, intended use, interfaces, controls and retained records. Add one failure signal and one person responsible for checking it. Keep this limit visible: A vendor brochure does not describe the deployed configuration or the study-specific data flow.
A strong science note separates observation from interpretation. Write a system card with name, version, owner, intended use, interfaces, controls and retained records. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Write a system card with name, version, owner, intended use, interfaces, controls and retained records. Keep this boundary visible: A vendor brochure does not describe the deployed configuration or the study-specific data flow.
CHAPTER 14 OF 17 · Preserve the trail
14. Retention should survive system and vendor change
Required records must remain complete, accessible and readable for the applicable retention period. Export formats, metadata, audit trails, certified copies and ownership agreements matter when a site changes vendors, software is retired or a sponsor relationship ends.
For retention should survive system and vendor change, provenance is the scientific story. Required records must remain complete, accessible and readable for the applicable retention period. Export formats, metadata, audit trails, certified copies and ownership agreements matter when a site changes vendors, software is retired or a sponsor relationship ends. That story remains bounded because Keeping a file is insufficient when the context or software needed to interpret it has vanished.
Try this
Turn retention should survive system and vendor change into a before-and-after audit. Test whether an independent reviewer can open, read and trace the retained record without the original vendor. Explain what evidence would reveal an incorrect transfer or correction. Respect this boundary: Keeping a file is insufficient when the context or software needed to interpret it has vanished.
A strong science note separates observation from interpretation. Test whether an independent reviewer can open, read and trace the retained record without the original vendor. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Test whether an independent reviewer can open, read and trace the retained record without the original vendor. Keep this boundary visible: Keeping a file is insufficient when the context or software needed to interpret it has vanished.
CHAPTER 15 OF 17 · Preserve the trail
15. FDA inspection needs direct, understandable access
FDA must be able to inspect source data and reconstruct the investigation. Access arrangements should protect confidentiality while providing authorised inspectors with records, metadata and audit trails in a usable form. Study teams should know who enables access and how delayed or remote access will be handled.
A useful test of fda inspection needs direct, understandable access is whether a new reviewer can reconstruct it without asking the original team. FDA must be able to inspect source data and reconstruct the investigation. Access arrangements should protect confidentiality while providing authorised inspectors with records, metadata and audit trails in a usable form. Study teams should know who enables access and how delayed or remote access will be handled. Preserve this caution: Security restrictions should not make required records effectively unavailable for inspection.
Try this
Teach fda inspection needs direct, understandable access using claim, source, transformation and limitation. Run an inspection drill for one participant record, including access approval, retrieval, audit trail and export. Let a partner challenge the weak link. Do not lose this guardrail: Security restrictions should not make required records effectively unavailable for inspection.
A strong science note separates observation from interpretation. Run an inspection drill for one participant record, including access approval, retrieval, audit trail and export. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Run an inspection drill for one participant record, including access approval, retrieval, audit trail and export. Keep this boundary visible: Security restrictions should not make required records effectively unavailable for inspection.
CHAPTER 16 OF 17 · Preserve the trail
16. eSource does not erase participant privacy duties
Electronic capture can concentrate detailed clinical information and expand the number of systems or service providers handling it. Role-based access, secure transfer, minimum necessary views, incident response and clear consent information help protect confidentiality without hiding legitimate inspection and monitoring access.
Follow esource does not erase participant privacy duties like a laboratory observation travelling through a network. Electronic capture can concentrate detailed clinical information and expand the number of systems or service providers handling it. Role-based access, secure transfer, minimum necessary views, incident response and clear consent information help protect confidentiality without hiding legitimate inspection and monitoring access. The trustworthy conclusion keeps this limit beside it: Strong privacy controls do not establish that the clinical measurement itself is accurate.
Try this
Build a provenance card for esource does not erase participant privacy duties. Draw the data-access map and distinguish privacy controls from measurement-quality controls. Add one failure signal and one person responsible for checking it. Keep this limit visible: Strong privacy controls do not establish that the clinical measurement itself is accurate.
A strong science note separates observation from interpretation. Draw the data-access map and distinguish privacy controls from measurement-quality controls. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Draw the data-access map and distinguish privacy controls from measurement-quality controls. Keep this boundary visible: Strong privacy controls do not establish that the clinical measurement itself is accurate.
CHAPTER 17 OF 17 · Preserve the trail
17. A complete eSource receipt makes the trial reproducible
A strong receipt records the protocol field, originator, original system, direct-entry or transfer route, identifier, timestamp, units, validation, review, corrections, queries, signatures, retention and inspection path. It allows a new reviewer to follow the value from observation to analysis without guessing.
For a complete esource receipt makes the trial reproducible, provenance is the scientific story. A strong receipt records the protocol field, originator, original system, direct-entry or transfer route, identifier, timestamp, units, validation, review, corrections, queries, signatures, retention and inspection path. It allows a new reviewer to follow the value from observation to analysis without guessing. That story remains bounded because A complete trail proves provenance; it does not prove that the protocol asked the right scientific question.
Try this
Turn a complete esource receipt makes the trial reproducible into a before-and-after audit. Hand the receipt to a classmate and ask them to reconstruct the datum, every change and its final analytical use. Explain what evidence would reveal an incorrect transfer or correction. Respect this boundary: A complete trail proves provenance; it does not prove that the protocol asked the right scientific question.
A strong science note separates observation from interpretation. Hand the receipt to a classmate and ask them to reconstruct the datum, every change and its final analytical use. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Hand the receipt to a classmate and ask them to reconstruct the datum, every change and its final analytical use. Keep this boundary visible: A complete trail proves provenance; it does not prove that the protocol asked the right scientific question.
eSource provenance map
| Question | Evidence to capture | Boundary to retain |
|---|---|---|
| Where did the datum begin? | Original record, originator, time, unit and identifier | Later entry is not the original observation |
| How did it reach the eCRF? | Direct entry, interface, mapping and validation | Automation can scale a mapping error |
| Who reviewed or corrected it? | Queries, old/new values, reason, review and signature | Sponsor review should not rewrite site source |
| Can it be reconstructed later? | Retention, metadata, audit trail, export and inspection | A file without context may be unusable |
Science learning becomes useful when a student can show where a claim came from, why it applies and where it stops. Continue with the eduKate Science Learning Hub, compare choices in the Education Hub, and connect these habits to career pathways built from subject capabilities.
Good eSource practice lets science move faster without losing its memory. A well-labelled datum can travel through people, devices and systems while still answering the simplest and most important question: where did this number come from?
Primary sources and verification date
Source status checked 26 October 2026. FDA guidance describes the agency’s current thinking and does not replace applicable statutes, regulations or product-specific communication with FDA.
- FDA: Electronic Source Data in Clinical Investigations — Final guidance, September 2013; docket FDA-2010-D-0643.
- FDA: Full eSource guidance PDF — Official recommendations on originators, identifiers, eCRFs, review and retention.
- FDA: E6(R3) Good Clinical Practice — Final Level 1 guidance, September 2025; current quality and data-governance context.