Trials which stop earlier than expected

Trials have a planned end goal. They are planned to end when

  • The required numbers (for study power) are enrolled
  • The required event count is achieved
  • The required follow-up period elapses for the enrolled patients
  • The planned end-of-trial date arrives (This has much to do with funding and logistics)

As data piles up, interim analysis is carried out.

This short of analysis should be performed according to certain validity-preserving criteria:

  • Planned a priori, rather than embarked upon when interesting things are observed
  • Carried out by an independent body, unconnected to the trial
  • Should be subject to correction which acknowledges the error-amplifying effects of multiple interim analysis.

Stopping a trial early - i.e. before the above goals are reached - can degrade the validity of the trial.
So why might a trial stop early?

  • Interim analysis demonstrates clear evidence of harm: the patients in the study group are dying because of the intervention, and it is unethical to continue administering it.
  • Interim analysis demonstrates clear evidence of benefit: the patients in the control group are dying for lack of effective treatment, and it is unethical to withold the study treatment.
  • Interim analysis points to futility: the study intervention is equivalent to placebo and continuing with the study will not increase the statisitical significance of the effect.
  • Trial is unviable: either it has run out of money, or the patients are too rare and recruitment is difficult, or the trial has taken too long and changing pracice has invalidated data which were collected earlier.

Why does it matter that a trial has topped early?

  • The reasons for stopping mst be clearly reported.
  • If the reason for stopping was unrelated to the data, then the conclusions are probably still valid
  • If the reason for stopping was something data-related, then perhaps the incomplete data set offers an incomplete picture, and the trial conclusions cannot be valid.

LITFL offers a couple of links to (Consolidated Standards for Reporting Trials), which is a set of standards and analytical checklists in support of transparency in the reporting of clinical trials. CONSORT have a good statement on the various (validity-preserving and invalidating) reasons for why a trial might be stopped.