CliniGA CRO

CliniGA CRO Blog

Duplicate Transfers: Avoiding New Records from Repeated Deliveries

Test receipt and reprocessing rules so that resending a clinical data file does not create duplicate observations.

Representative research photograph: Duplicate Transfers: Avoiding New Records from Repeated Deliveries
Photo: CDC Global · License: CC BY 2.0 · Cropped and resized.
Published
Updated

Resending a file after a network failure is normal. If a receiver treats repeated measurements as new observations, a technical recovery action can create duplicate records in the clinical dataset without attracting immediate attention.

Preserve the meaning of data

Distinguish file identity, record identity and correction version during interface design. Repeated delivery is different from a genuinely new measurement for the same participant. Define the expected result of reprocessing explicitly. A repeat-resistant acceptance process must also avoid discarding valid corrections or legitimate new data simply because some identifiers are familiar.

Worked example

For example, a laboratory file is received successfully but the acknowledgement never reaches the sender. The sender retries. Here, the content fingerprint and transaction identity are compared with the earlier acceptance record, and the repeated delivery creates no new results. A corrected file is then tested separately to ensure that the same protection does not block an authorised update.

Review checks

  • Resend an identical file consecutively and after an interrupted session.
  • Test the key that distinguishes a correction from a delivery retry.
  • Review records of rejected repeats and accepted corrections together.

Acceptance and boundaries

Acceptance is demonstrated when repeated delivery preserves the expected record count and version relationships. A content fingerprint alone does not prove clinical correctness; it helps recognise identical bytes. Changing the rules requires a controlled decision about historical reprocessing and renewed reconciliation. Keep enough delivery evidence to explain why each incoming transaction was accepted or skipped.

Frequently asked questions

Should every repeated file be deleted?
Not necessarily. Delivery evidence can be retained while preventing the same observation from becoming a second valid result.
Does a different file name mean new data?
No. Different names may still represent the same content and business transaction.

Sources

  1. FDA — Electronic systems, records and signatures in clinical investigations
  2. MHRA — Guidance on GxP data integrity