THE STATISTICAL ATTRIBUTION OF CAUSATION Prof. Deborah Ashby, London, United Kingdom    - Biography English - 2007-10-08

PLAY PRESENTATION	( 30 slide(s) )
PODCAST

VIEW THE SLIDE LIBRARY FOR THIS PRESENTATION

Background and Aims
When using medicines for children, we need to know about their efficacy and safety. Typically evidence on efficacy comes from randomised controlled trials, which are fairly straightforward to interpret, at least when they have been carried out in relevant populations. However data on safety can come from a variety of sources, and so interpretation is much more challenging. The aim of this talk is to explore the extent to which assessment of causality is possible when faced with observational evidence.

Methods
This talk will outline the various sources of data, from anecdotal case reports to designed observational studies and randomised controlled trials, together with their strengths and weaknesses. Austin Bradford-Hill proposed criteria for assessment of causality of observed associations: temporal association, strength of association, a dose-response relationship, consistency, plausibility, existence of alternative explanations, experimental evidence, specificity and coherence of the evidence. Their application in this context will be explored.

Results
We will look in detail at three questions: ‘Do high-strength pancreatic enzymes cause fibrosing colonopathy in children with cystic fibrosis?’ ‘Do SSRI anti-depressants cause suicidal behaviour in children?’ and ‘What dose of folic acid should be used to lower the risk of neural tube defects?’

Conclusions
The assessment of the safety of medicines depends upon the integration of evidence from various sources. Used carefully and intelligently, such evidence can help make medicines safer for children.