Causality, drug-effect assessment

Temporal Relationships of Adverse Events. The temporal relationship between duration of product exposure and development of an adverse event is important in assessing causality. But how can data on temporal relationships be systematically summarized in a database containing thousands or even hundreds of thousands of subjects Temporal relationships cannot be clearly elicited if only frequencies of adverse events between treatment and control groups are compared. There can be many disparities in the subjects time of exposure or time at risk. Toxic manifestations of drugs may not occur until several months or even years after the initial exposure to the drug. How do we systematically assess delayed toxicity of a previously prescribed drug from the effect of a newly prescribed drug Such a scenario occurred with reported cases of pancreatitis associated with valproic acid therapy, in which some cases appeared several years after therapy [2]. 

This is the classic randomised controlled trial (RCT), the most secure method for drawing a causal inference about the effects of treatments. Randomisation attempts to control biases of various kinds when assessing the effects of treatments. RCTs are employed at all phases of drug development and in the various types and designs of trials discussed below. 

A major source of confusion in the media relates to how science tries to separate unavoidable death due to natural causes from avoidable death due to modem drugs or chemicals. This is the field of risk assessment and is a crucial function of our regulatory agencies. Confusion and unnecessary anxiety arise from the complexity of the underlying science, misinterpretation of the meaning of scientific results, and a lack of understanding of the basic principles of probability, statistics, and causal inference (assigning cause and effect to an event). 

No formal assessment of the validity of each reference was undertaken in this process, although the levels of evidence afforded by different types of publications (i.e., case report vs. randomized, placebo-controlled double-blind study) were actively considered during the review process. In addition, it was observed that some identified publications were of limited value, especially those that lack sufficient detail about the specific herbal preparation addressed, and case reports that postulate a causal relationship between a specific herbal ingredient and a reported adverse effect, without consideration for confounding factors such as patient history or concomitant drug use. Some such references were nonetheless retained, though the editors attempted to call attention to their perceived flaws. 

---