Case Study 2

Client

Big Pharma company

Objective

Anonymization for Project for Policy 0070 submission

Project Overview

  • Complete anonymization with generation of anonymization report for 2 Studies.
  • 12 and 14 analysis datasets in respective studies
  • Approximately 5000 pages to anonymize

Challenges

  • Received project data in last week of June 2018 with timeline 
  • to complete in 2 weeks
  • Due to Brexit, EMA was accepting submission of redaction proposal only till end of July’18 and final package till end 
  • of August’18.

What we delivered

  • Complete anonymization of the 2 studies with generation 
  • of anonymization report and finalized redaction package before due date.
  • Anonymized documents published on EMA website in Sep 2018

Project Highlights

  • De-identified dataset initially as per PhUSE rules, then adjusted de-identification to achieve residual risk threshold. Methods used were:

º Pseudonymize

º Date Offset

º Generalization for Age, Country, Race, Ethnicity

º Variable suppression for free text variables

º De-identified values from dataset to document were done using SHADOW™ tool

º Risk of re-identification in CSR was based on complete set of patients in project.

º Link between data is maintained between analysis datasets and CSR documents

º Risk of (attempt from adversary) = 1 in lieu of public release.

º Risk of re-identification calculated for different combinations of quasi-identifiers as suggested in policy 0070, Kniola (2016), IOM guidelines.

º Risk Threshold = 0.09

º Risk metric used = Maximum

º Risk of re-identification in de-identified data= 0.0286

geninvo logo

Ideas Delivered Intuitively