Visual Analytic Approaches for the Analysis of Spontaneously Reported Adverse Events in Post-Market Surveillance

572 views

Published on

This presentation was given live at JMP Discovery Summit 2012 in Cary, North Carolina, USA.More information on JMP Clinical is available at http://www.jmp.com/software/clinical/

Published in: Technology
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
572
On SlideShare
0
From Embeds
0
Number of Embeds
0
Actions
Shares
0
Downloads
7
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Visual Analytic Approaches for the Analysis of Spontaneously Reported Adverse Events in Post-Market Surveillance

  1. 1. Copyright © 2010 SAS Institute Inc. All rights reserved. Visual Analytic Approaches for the Analysis of Spontaneously-Reported Adverse Events in Post-Market Surveillance Richard C. Zink, Ph.D. Principal Research Statistician Developer JMP Life Sciences richard.zink@jmp.com
  2. 2. 2 Copyright © 2010, SAS Institute Inc. All rights reserved. Safety Analyses  Randomized clinical trials are the gold standard for evaluating the efficacy of a new intervention  Development program provides an incomplete safety profile  Numerous safety endpoints (AE, labs, vitals, hospitalizations, QOL, ECG; multiplicity)  Rare safety events require greater sample size (power)  Studied in limited population » Those most likely to respond for efficacy » Those on few or no medications with limited other confounding diseases  Animals models may not be predictive (DILI)  Limited understanding of biological mechanisms and pathways
  3. 3. 3 Copyright © 2010, SAS Institute Inc. All rights reserved. Safety Recalls  Vioxx (rofecoxib). COX-2 inhibitors have increased CV risk and life-threatening GI bleeding  Avandia (rosiglitazone). Increased CV risk.  Rezulin (troglitazone). Liver failure  Tysabri (natalizumab). Increased risk of PML (progressive multifocal leukoencephalopathy)  Risk a therapy may pose may not be well understood until it has been on the market for many years, taken by individuals who differ from those studied under the inclusion criteria of the clinical development program
  4. 4. 4 Copyright © 2010, SAS Institute Inc. All rights reserved. Spontaneously-reported adverse events  Collected by regulatory agencies, pharmaceutical companies and device manufacturers to monitor the safety of a product once it reaches the market  FDA maintains  AERS: Adverse Event Reporting System  VAERS: Vaccines Adverse Event Reporting System  MAUDE: Manufacturer and User Facility Device Experience  Data are obtained from physicians, patients or medical literature
  5. 5. 5 Copyright © 2010, SAS Institute Inc. All rights reserved. Data Structure Case Event 1 Event 2 Event 3 Drug 1 Drug 2 Drug 1 Drug 1 Drug 2 Drug 3 Figure 1. Data Structure One record (row) per case-event-drug in a SAS formatted data set. This example would have 6 rows.
  6. 6. 6 Copyright © 2010, SAS Institute Inc. All rights reserved. Spontaneously-reported adverse events  Unique challenges  No measure of total exposure (total number of subjects taking the drug)  No “control”  Data often inconsistently captured and of poor quality  Of reported events, is a drug-event combination occurring more often than expected assuming independence between drug and events?  Important:  No guarantee drug caused event  Number of cases not equal to number of people. Can get multiple reports from same person  Cannot calculate incidence
  7. 7. 7 Copyright © 2010, SAS Institute Inc. All rights reserved. Pharmacovigilance: Disproportionality Analysis Event of interest Drug of interest Mentioned Not mentioned Mentioned 𝑛ℎi𝑗 𝑛ℎ𝑖. − 𝑛ℎi𝑗 𝑛ℎ𝑖. Not mentioned 𝑛ℎ.𝑗 − 𝑛ℎi𝑗 𝑁ℎ − 𝑛ℎ𝑖. − 𝑛ℎ.𝑗 + 𝑛ℎi𝑗 𝑁ℎ − 𝑛ℎ𝑖. 𝑛ℎ.𝑗 𝑁ℎ − 𝑛ℎ.𝑗 𝑁ℎ Table 1. Contingency Table for Drug i and Event j in Stratum h Nh is the number of cases in stratum h. Use the above contingency table to define measures of disproportionality 1. Reporting Odds Ratio (Meyboom et al., 1997) 2. Proportional Reporting Ratio (Evans, Waller & Davis, 2001) 3. Multi-item Gamma Poisson Shrinker (DuMouchel, 1999) 4. Bayesian Confidence Propagation Neural Network (Bate et al., 1998; Gould, 2003). Stratification is an important consideration for disproportionality analysis. How drugs are prescribed, disease severity and how individuals may respond to treatment can all be influenced by demographic and other background characteristics. Therefore, disproportionality statistics should be calculated among homogeneous groups to avoid inappropriate conclusions (Woo et al., 2008).
  8. 8. 8 Copyright © 2010, SAS Institute Inc. All rights reserved. Pharmacovigilance: Disproportionality Analysis  How is a signal defined?  Each drug-event pair has a confidence or credible interval associated with it  If this interval exceeds 1, it is an indication that the event occurs more often for the drug  May want a higher threshold  May include frequency of events

×