One day, I validated an AESI(adverse event of Special Interest) data, I found a discrepancy between two output datasets. After I notified my boss, she suggested checking the AESI excel sheet, and it turns out both versions are partially correct. Because the Prefered term Hepatic Failure belongs to both Hepatic Disorder and Hepatic Failure EOI in a particular study. Instead of mapping it to a single EOI, we should map it to both.
Today we’ll go over the basics of AESI and how to deal with the situation. One MedDRA Prefer term maps to multiple EOI(Event of Interests) terms.
AESI: An adverse event of special interest is one of scientific and medical concern specific to the sponsor’s product or program.
How to generate AESI:
Extract the study-based AESI info from EXCEL(EOI, SMQ, CQxxnam).
Merge AESI with AE dataset by Prefered term(aedecod)
Set AESI related Flag
Adjudication process with the medical team and cross-check with CRF
Generate AESI related output
Sample Code for single Prefered term map to Multiple EOI(Event of Interest)
data eoi1; set eoi; eoi="EOI/Search Strategy"n; aedecod="Meddra pt"n; aeptcd="Meddra code"n; proc sort nodupkey; by aedecod eoi aeptcd; run; *mutliple eoi; proc transpose data=eoi1 out=eoi1t; var eoi; by aedecod aeptcd; run; data eoi2; set eoi1t; eoic=catx(";", strip(col1), strip(col2)) /*coln as needed*/; proc sort; by aeptcd aedecod; data ae1; merge eoi2(in=a) ae(in=b); by aeptcd aedecod; if b; run;
Thanks Helen for pointing out the discrepancy that might occur due to the one-to-many mapping. It really made me think when differences happen, instead of focusing on what others did wrong, maybe we all did partially correct!