Download presentation
Presentation is loading. Please wait.
Published byAntony Chandler Modified over 9 years ago
1
Screening for carcinogenic or chemopreventive effects of prescription drugs Anton Pottegård, MScPharm PhD University of Southern Denmark
3
VKA – All cancers Benzodiazepiner – All cancers Azathioprine – All cancers [Myasthenia] Disulfiram – MM/Breast/Prostate [Alcoholics] Azathioprine – NMSC [Myasthenia] Myasthenia – All cancers Metformin – Colon cancer [Diabetics] Statins – NMSC Lithium – Upper urinary tract cancer Statins – Renal cancer Lithium – Colorectal cancer PDEIs – Malignant melanoma TCA – Glioma
4
RACED Rapid Assessment of Carcinogenic Effects of Drugs
9
RACED vol. 1 Table of contents: Valproic acid (sub. prostate) Allopurinol Antidepressants Antipsychotics (sub. prostate) Ibuprofen (sub. colon) Thiazides Ace-inhibitors ATII-antagonists Calcium-channel-blockers (grp2) Clopidogrel
13
Dicloxacillin Pottegård et al. JAMA 2015 Jul 21;314(3):296-7. Proton Pump Inhibitors Henriksen et al. Pharmacoepidemiol Drug Saf. 2015 Sep 23. Antidiabetics Stage et al. J Thromb Haemost (Accepted)
15
RACED2 Anton Pottegård Søren Friis René dePont Christensen Laurel A. Habel Joshua J. Gagne Jesper Hallas
16
Design Multiple case-control study
17
Outcome All cancer Single cancer sites Single cancer types Based on ICD-O topography and morphology codes: 99 cancers and subtypes.
19
Exposures Only drug classes Only single substances Both! For each cancer, we included all drugs that had 10 observed long-term users (defined as ≥8 prescriptions) among the cases
20
Analysis Matching by gender and age … thereby adjusting for gender, age and calendar time Adjusting for Charlson-score and education … as crude confounder adjustment
21
Material 278,485 incident cancer cases 22,125 drug-cancer pairs
22
Screening: Step 1 ORs estimated comparing long-term use (≥8 fills) to non-use. Signals were defined as: OR 1.5, or 95%CI >1.2 or <0.83
23
Intermediary results 4,561 signals
24
Screening: Step 2 Outcome specificity The ratio of OR signal to OR overall outside 0.83-1.20. Dose-response pattern A significant slope on the dose-response curve (p<0.10).
25
Results 77.4% failed to meet dose-response criteria, 1.7% failed test for specificity, 1.4% failed both; thus leaving 1,020 signals. 159 at the second level of the ATC-system, 351 at the fourth level of the ATC-system, and 510 for single agents. 703 signals indicated an increased cancer risk, 317 indicated a chemopreventive effect.
29
All done? What about issues of multiple testing? What about legal issues?
32
Male pharmacoepidemiologist seeks enthusiastic partner who owns his/her own degree and grows his/her own hypotheses. Access to cancer-data is a plus, as is being open-minded!
33
Anton Pottegård apottegaard@health.sdu.dk www.antonpottegaard.dk 33
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.