The afternoon is a different story. Participant three arrived 20 minutes late, which compressed the visit into the lunch hour. The coordinator took vital signs, performed the assessments, dispensed the investigational product, and moved directly to participant four, who was already waiting. Now, nearly three hours after seeing participant three and 90 minutes after participant four, the coordinator sits down to reconstruct what happened.

The blood pressure for participant three -- was it 138/86 or 136/88? Both numbers feel right. The coordinator wrote the systolic on the corner of the visit schedule but did not note the diastolic. The adverse event that participant four mentioned -- was it the headache that started Tuesday or Wednesday? She said the word "Tuesday," but then corrected herself, or maybe she was talking about a different symptom when she said that. And did the coordinator record that participant three denied any new medications, or did that conversation happen with participant four?

This is the moment where data integrity begins to erode. Not because the coordinator is careless. Not because the coordinator does not understand the protocol. But because human memory is a reconstructive process, not a recording device, and three hours is more than enough time for details to shift, merge, and fabricate themselves with absolute conviction.

The afternoon is a different story. Participant three arrived 20 minutes late, which compressed the visit into the lunch hour. The coordinator took vital signs, performed the assessments, dispensed the investigational product, and moved directly to participant four, who was already waiting. Now, nearly three hours after seeing participant three and 90 minutes after participant four, the coordinator sits down to reconstruct what happened.

The blood pressure for participant three -- was it 138/86 or 136/88? Both numbers feel right. The coordinator wrote the systolic on the corner of the visit schedule but did not note the diastolic. The adverse event that participant four mentioned -- was it the headache that started Tuesday or Wednesday? She said the word "Tuesday," but then corrected herself, or maybe she was talking about a different symptom when she said that. And did the coordinator record that participant three denied any new medications, or did that conversation happen with participant four?

This is the moment where data integrity begins to erode. Not because the coordinator is careless. Not because the coordinator does not understand the protocol. But because human memory is a reconstructive process, not a recording device, and three hours is more than enough time for details to shift, merge, and fabricate themselves with absolute conviction.