A licensing decision made under uncertainty has to remain revisable

The argument

A second argument offered for this position concerns what kind of act an approval is. Proponents describe it as a judgement made under acknowledged uncertainty: a regulator weighs an incomplete picture of benefit against an incomplete picture of harm and decides that, on the evidence available, the balance favours allowing the medicine to be sold. Every part of that sentence is provisional. The evidence is partial, the comparison is against whatever alternatives existed at the time, and the population the medicine will be used in is not yet the population it was studied in. The argument is that a decision of this kind is only defensible if it can be revisited, and that revisiting it requires machinery. Someone has to notice that a pattern of reports has exceeded what background rates would produce, decide whether the pattern is causal, and then act on it by changing a label, narrowing an indication, adding a contraindication, restricting prescribing to specialists or removing the medicine altogether. Proponents point out that none of this happens by default. Without a system built to do it, an approval granted once stands indefinitely, not because it has been confirmed but because nothing exists to overturn it, and the uncertainty acknowledged at the moment of licensing quietly becomes a settled fact. Holders of the view add that this changes what a regulator can safely approve. A system that can withdraw and amend is able to license a promising medicine on thinner evidence, because the decision is correctable. A system that cannot must either demand evidence that takes years to assemble or accept consequences it has no way to reverse. Surveillance on this account is what makes the rest of regulation possible, since it is the only part that can change a decision after the fact.

Premises

Counter-arguments

Opponents accept that approvals must be revisable and deny that this makes surveillance the priority. The capacity to amend a licence, they argue, is a legal and administrative power rather than a scientific programme, and it can be held in reserve cheaply while resources go to the stage that determines whether the decision was sound in the first place. Several add that the argument proves too much, because if approvals can always be corrected later then the standard of evidence required to grant one tends to fall, and a system that licenses on thinner evidence precisely because it can withdraw afterwards has shifted the cost of its uncertainty onto the patients who take the medicine in the interval.

Rejecting the premises

[Rejecting P2] The power to amend or revoke an authorisation is largely legal and administrative, and need not be organised as the centre of a regulatory system in order to be available. [Rejecting C] Treating correctability as the priority tends to lower the evidence required for approval, which shifts the cost of the remaining uncertainty onto patients treated before any signal emerges.