Background: Patient expectations may be an important component of a placebo effect, and yet few studies assess how patients' pre-existing expectations affect subsequent health outcomes.
Objectives: To estimate the association between patients' expectations and time to cure in patients with clinically diagnosed acute bacterial rhinosinusitis.
Methods: In a randomised controlled trial, expectations about the benefit of an antibiotic therapy were measured prior to treatment with either an antibiotic or placebo.
Results: Of the patients asked for consent, 64% refused mostly because they either wanted or did not want to receive antibiotics. Over 25% of the patients who gave consent were ambivalent about the benefit of antibiotic therapy. Predictably there was no evidence of an association between expectations and time to cure in those that gave consent.
Conclusions: Selection bias occurs if patients with strong expectations refuse to participate in a trial and if the success of the intervention depends in part on a patient's expectations. Methods that adjust for selection bias are recommended for trials where placebo effects are either of interest or could be an important component of an intervention.