Statistical analysis of possible bias of clinical judgements due to observing an on-therapy marker variable

Preventing febrile seizures in children with oral diazepam: can a controlled trial truly be "double-blind?"

OBJECTIVE

To determine in a randomized, double-blind, clinical drug trial in children whether parental "blindness" is maintained.

STUDY DESIGN

Oral diazepam or placebo was given for fevers to 406 children with at least one previous febrile seizure. Later, 192 of these families (102 diazepam, 90 placebo) were contacted and asked: (1) Did you give your child the study medicine for fevers? (2) Do you think you knew your child's treatment group (diazepam or placebo)? (3) If you think you knew, why?

RESULTS

In the group of children randomly assigned to receive diazepam, 69% of their parents guessed correctly. In the group assigned to receive placebo, only 19% of parents guessed correctly. Parental opinion was influenced mostly by the presence or absence of side effects, and treatment efficacy or failure was the next most important factor.

CONCLUSION

Because in a double-blind clinical trial, many parents can correctly guess that their child is receiving active drug, this may influence compliance with the protocol. Thus safeguards are needed to reduce parental bias that can invalidate the results of double-blind clinical trials.