Tuesday, March 24, 2015

Randomized Control Trials

Nancy Cartwright, who is a philosopher of science, wrote this interesting article in the Lancet (open-access) called "A philosopher's view of the long road from RCTs to effectiveness".

In RCTs, eligible individuals are randomly assigned to a control group (which does not receive the intervention), and a treatment group (which does). The main advantage is that RCTs "control for unknown confounders", and any significant difference in outcome between the control and treatment groups, in an ideal RCT, may be causally ascribed to the intervention.

Some of the well-known advantages, successes, and disadvantages are listed in wikipedia's entry on the topic.

Dr. Cartwright describes the premises that need to be in place for an RCT to be "ideal", and then goes into some more philosophical limitations:
[RCTs] are ideal for ensuring “the treatment caused the outcome in some members of the study”—ie, they are ideal for supporting “it-works-somewhere” claims. [...] For policy and practice we do not need to know “it works somewhere”. We need evidence for “it-will-work-for-us” claims: the treatment will produce the desired outcome in our situation as implemented there.
She offers some suggestions. Definitely worth a look.

No comments: