Identification errors in medical research: privacy at all costs?

In laboratory medicine, a misidentified patient sample can lead to an incorrect tissue diagnosis, a potentially fatal blood transfusion error or other serious adverse events. Although well characterised in routine patient care, the overall impacts of misidentification errors in the clinical research setting are less conspicuous but potentially greater, with downstream effects that may extend beyond care at an individual level. When data discrepancies or queries arise in clinical trial data then a data clarification form (DCF) is issued to the researcher by the overseeing trial coordinator or sponsor. Higher rates of DCF's are sometimes used as a crude surrogate marker of poorer trial quality. However, data is scarce on misidentification rates in clinical trials. In five clinical trials involving 822 histology or blood specimens analysed by our pathology department, DCF's were issued for 21% (174) of specimens. Amongst these 67% (117 / 174) were related to sample identification. Although these errors were recognised before data was compromised or an adverse event occurred, they highlight an alarming lack of stringency of use of patient identifiers in the research setting. We therefore propose the use of an appropriate number of de-identified data points and a formalised specimen accession process as employed in routine care to mitigate misidentification errors and their impact in clinical research. Increased recognition in the research community of the likely effect of truncating or reducing the number of patient identifiers is needed to minimise misidentification errors in the research setting.