Causal inference in biomedical research
- PDF / 553,319 Bytes
- 19 Pages / 439.37 x 666.142 pts Page_size
- 59 Downloads / 265 Views
Causal inference in biomedical research Tudor M. Baetu1 Received: 10 September 2019 / Accepted: 15 July 2020 © Springer Nature B.V. 2020
Abstract Current debates surrounding the virtues and shortcomings of randomization are symptomatic of a lack of appreciation of the fact that causation can be inferred by two distinct inference methods, each requiring its own, specific experimental design. There is a non-statistical type of inference associated with controlled experiments in basic biomedical research; and a statistical variety associated with randomized controlled trials in clinical research. I argue that the main difference between the two hinges on the satisfaction of the comparability requirement, which is in turn dictated by the nature of the objects of study, namely homogeneous or heterogeneous populations of biological systems. Among other things, this entails that the objection according to which randomized experiments fail to provide better evidence for causation because randomization cannot guarantee comparability is mistaken. Keywords Randomization · Controlled experiment · RCT · Experimental science
A debate concerning the virtues of randomization Clinical trials aim to determine whether a medical intervention is a causal difference maker in respect to a health-related outcome in a patient or, more often, a population of patients. Randomized Controlled Trials (RCTs) are widely regarded as the gold standard in clinical research, providing the strongest evidence for causal efficacy (The Cochrane Collaboration 2011). It is not surprising, therefore, that one of the most debated questions is how exactly and to what extent the main feature that differentiates RCTs from other types of controlled experiments, namely the random allocation of subjects to the test and control arms of the experiment, contributes to the validity of causal inference. A common answer is that randomization ensures comparability, that is, an even distribution of potential confounders among patients in the test and control * Tudor M. Baetu tudor‑[email protected] 1
Département de philosophie et des arts, Université du Québec à Trois-Rivières, 3351, boul. des Forges, Trois Rivières, Québec G8Z 4M3, Canada
13
Vol.:(0123456789)
43 Page 2 of 19 T. M. Baetu
groups (Cartwright 2010; Papineau 1994). Critics challenge this claim, pointing out that randomization can balance the effects of confounders only in the long run, by performing an infinite series of experiments in which patients are randomly allocated to test and control conditions (Howson and Urbach 2006; Lindley 1982; Urbach 1985, 1993; Worrall 2007b). It would seem therefore that critics too, assume that valid statistical inference in clinical research requires comparable groups balanced in respect to potential confounders and that the main purpose of random allocation is to achieve comparability. However, since randomization cannot guarantee comparability, they conclude that randomized studies are not epistemically superior to non-randomized ones. But why is compara
Data Loading...