Virginia Falcón Advisor: Prof. Prashant Yadav

US $2.5 billions in annual sales lawsuits May 1999: FDA approves Vioxx March 2000: study reveals that Vioxx patients are 2 times more likely to have cardiovascular problems than Naproxen April 2002: The FDA tells Merck to add information about cardiovascular risk to Vioxx's label. Aug. 25, 2004: database analysis of 1.4 million patients reveals that Vioxx users are more likely to suffer cardiac problems than those taking Celebrex. Sept. 23, 2004: Merck says it learned this day that patients taking Vioxx in a study were twice as likely to suffer a heart attack or stroke as those on placebo. Sept. 30, 2004: Merck withdraws Vioxx from the U.S. and the more than 80 other countries in which it was marketed. Pharmaceuticals Big trouble for Merck Nov 4th 2004 Merck Agrees to Settle Vioxx Suits for $4.85 Billion By ALEX BERENSONALEX BERENSON Merck recalls Vioxx Merck Admits a Data Error on Vioxx Arthritis drug Vioxx yanked off market Study by the maker confirms risk of heart attack, stroke Company's study confirms risk of heart attack, stroke Sabin Russell, Chronicle Medical Writer Friday, October 1, 2004

 Defined by the WHO as “the science and activities relating to the detection, assessment, understanding and prevention of adverse effects or any other possible drug- related problems”  Adverse Drug Reactions account for around 3% of deaths in the developed world.

Pattern of disease burden and consequently drug use are different (developed countries minimally use anti-TB, anti-malarial and anti-diarrheal drugs, for example) Patient genotype and phenotype are markedly distinct, with a large number of malnourished patients and patients with concomitant diseases in developing countries relative to developed countries Social and economic conditions are different (higher rates of poverty and illiteracy in developing countries) Formulations and brands of drugs used are different Diversity of drugs currently being launched (combination therapies) reaching developing countries in greater numbers and more quickly because of new funding Faster scale up of public health programs due to availability of new funding from major donors is exposing more patients to treatments.

Environment

 Determine proper cohort sizes for pharmacovigilance studies based on common sampling techniques.  Devise sampling strategies to select sentinel sites for data collection in order to reduce biases.  Propose techniques to compensate for biases.  Apply all of the above in a case study for Tanzania.

Select at the site level those sites which could potentially bring to the sample a representative mix of patients. Include in the design strategy the main sources of noise (confounding factors) so we can either block its effects. Three approaches were explored: Cluster, Stratified and Quota Sampling.

S S S S S S S S S S S S S S S S S S S S S S S S S S S S S - Variables: - Constraints: Minimum number of patients: Include Private Facilities: Include hospitals Male/Female Ratio Urban/Rural Ratio …

A A A A AB B B B BC C C C CD D D D DE E E E E F F F G G G G GH H H H HI I I I I I IJ J J J J J J K K K K KM M N N N N NO O O O OP P P Q Q Q Q QR R R R RS S S T T T T TU U U U U Clustering algorithms Integer Program Pool of Sites

- Compensation for non-reporters: - Cohort size: - Compensation for biases:

Stratified Sampling Cluster Sampling Quota Sampling

 Controlling for confounders at the site selection level for a CEM is both necessary and feasible.  Sampling techniques can be used for site selection.  The results of a study and its robustness could be greatly enriched by the use of this techniques.