Causality, Impartiality and Evidence-Based Policy - Mechanism and Causality in Biology and Economics - page 206

alone. Lots of formal and informal causal knowledge are acquired in the first two
phases, and it is put to test not only in the RCTs, but by the subsequent epidemio-
logical surveillance, if the drug is approved.
Expert judgement contributes to all the four phases, and even if the decision to
authorise a drug is taken at the third phase, the decision is not mechanical , precisely
because the external validity of a trial cannot be taken for granted. But this is
something pharmacologists have known right from the beginning: the regulatory
system at the FDA was established to deal with causal uncertainty. During most of
the second half of the twentieth century, pharmaceutical research advanced through
a so-called molecular lottery: compounds were synthesised and tested on animals
without any clear theoretical guidance, and even when they had an interesting
therapeutic effect, there often was no grasp of the precise mechanism responsible
for it. RCTs allowed pharmacologists to deal with this causal uncertainty about
drugs, focusing on the probability of attaining certain treatment outcomes under a
given range of prognostic factors.
But if RCTs rely so crucially on expert judgement, we may wonder why they
were considered an improvement over the older methodology. One of us has
defended elsewhere that all the involved parties explicit sought a testing methodol-
ogy with warrants of impartiality against the potential conflicts of interest arising in
the trial (Teira 2011a , b ). The 1950s saw a boom in industrial drug production
(some of which were 'wonder drugs' such as antibiotics, but many were just
combinations of already available compounds) and, simultaneously, in pharmaceu-
tical advertising that caused much confusion among practitioners about the thera-
peutic merit of each product (Marks 2000 , p. 346). For therapeutic reformers, RCTs
with their strict research protocol provided the information about drugs that 'sleazy
advertising' was trying to disguise with 'badly scissored quotes', 'pharmaceutical
numbers racket', 'detail men' visits and so forth (Lasagna 1959 , pp. 460-461).
Adopting RCTs as a regulatory tool allowed the FDA to justify the impartiality of
their decisions about treatments before patients, physicians and the pharmaceutical
industry (Teira 2011a , b ).
In this respect, randomisation was considered more as a debiasing procedure
than a tool for causal inference. Randomisation prevents researchers from
allocating treatments to patients according to their personal interests, so that the
healthiest patients get the researcher's favourite therapy. As mentioned above, such
unbalanced allocations can happen nonetheless by chance. But randomisation is
still a warrant that the allocation was not done on purpose with a view to promoting
somebody's interests. A priori, the experimental procedure is impartial with respect
to the interests at stake.
Of course, in Cartwright's causal approach, randomisation would just be a tool
for controlling the probabilistic dependences arising from selection biases in the
experimental and the control group, making sure that they are the same in both
treatment groups except for the treatment. To understand randomisation in this way
is perfectly appropriate from a methodological point of view, but this is not how it
was understood by those who introduced it into the regulatory system. Given how
little was known about causation in RCTs at the time, randomisation was not sought