Foundations of Evidence-Based Medicine

Pharmacology and Statistics: Recommendations to Strengthen a Productive Partnership

“…However, Carnap reinforced the notion that both these categories depended on the assumption of regularity in the future, i.e., that the past does, in fact, predict the future. In other words, we must assume that sound science can overcome Hume's 'problem of induction' [5,7,12] . Moreover, Carnap conceded that a continuum exists between universal laws and statistical laws [12] .…”

“…In recent publications, we and others have discussed truth, underlying assumptions in medical science, various notions of causality, general evidential criteria for establishing causality in the pharmacological sciences, and myriad challenges to these current principles, including Quine's correct notions of holism and underdetermination of theory with data [1,4,5,7,10] . We described how these challenges have been overcome, topics relevant to the current discussion [7] .…”

“…Much attention focused on discovering cause and effect relationships [5][6][7][8] . Since ultimate goals of pharmacology (encompassing both basic and clinical pharmacology) are prevention and treatment of disease and illness, pharmacological experiments must be performed, interpreted and reported correctly.…”

“…The concept of contributory causality requires statistical thinking and its attendant science since, for example, in treatment response, not everyone who receives a certain 'efficacious' treatment responds, nor does everyone not receiving the treatment fail to respond [7,12] . Much of causal thinking in pharmacology depends on the notion of contributory causality [2,[5][6][7]12] . Therefore, because of the nature of contributory causality, large randomized long-term clinical trials, especially when effects are modest or slow, are sometimes required to establish the value of certain treatments, e.g., lowering serum cholesterol with drugs [1,5,7,8,10] .…”

“…Much of causal thinking in pharmacology depends on the notion of contributory causality [2,[5][6][7]12] . Therefore, because of the nature of contributory causality, large randomized long-term clinical trials, especially when effects are modest or slow, are sometimes required to establish the value of certain treatments, e.g., lowering serum cholesterol with drugs [1,5,7,8,10] . Often, the objective of trialists is to falsify the null hypothesis so the treatment can be concluded to be 'better' than placebo (or another treatment) on the outcome variable or, in other words, that the treatment 'causes' improved health, at least as far as the primary outcome variable is concerned.…”