Endgames Statistical Question

Placebo run in periods

BMJ 2012; 344 doi: http://dx.doi.org/10.1136/bmj.e3827 (Published 07 June 2012) Cite this as: BMJ 2012;344:e3827
  1. Philip Sedgwick, senior lecturer in medical statistics
  1. 1Centre for Medical and Healthcare Education, St George’s, University of London, Tooting, London, UK
  1. p.sedgwick{at}sgul.ac.uk

Researchers investigated the efficacy of hypericum extract (St John’s wort) compared with paroxetine in patients with moderate to severe major depression. A randomised double blind, double dummy, multicentre non-inferiority trial study design of six weeks’ duration was used. Treatment consisted of 300 mg hypericum extract WS 5570 three times a day or 20 mg paroxetine once a day for six weeks. The primary outcome was change in score on the Hamilton depression scale from baseline at six weeks.1

In total, 301 patients were recruited from psychiatric primary centres in Germany. After a screening examination patients underwent a single (patient) blind, double dummy, placebo run in period of three to seven days before the trial started. Subsequently, 251 patients aged 18-70 years were recruited to the trial, and 125 were randomised to the hypericum treatment group and 126 to paroxetine. The researchers reported that in the treatment of moderate to severe major depression, hypericum extract WS 5570 was at least as effective as paroxetine and better tolerated.

Which of the following statements, if any, are true for the placebo run in period?

  • a) All participants received only placebo

  • b) Participants were randomised to hypericum placebo or paroxetine placebo

  • c) Participants did not know that they received a placebo


Statement a and c are true, whereas b is false.

All patients who volunteered for the trial entered the placebo run in period. The placebo run in period occurred before patients were randomised to treatment groups. During the placebo run in period patients received only placebo (a is true). In the example above, patients received a hypericum placebo and paroxetine placebo (b is false). The researchers reported that patients received three coated tablets of hypericum placebo a day plus one paroxetine placebo capsule in the morning.

Patients received hypericum placebo and paroxetine placebo in the placebo run in period because the trial required a double dummy design after randomisation. Blinding could not be easily achieved in the trial because the frequency of administration of hypericum and paroxetine differed, with hypericum three times a day and parotexine once in the morning. Blinding could be carried out in this trial after randomisation only by using a double dummy, which has been described in a previous question.2 This involved both treatment groups receiving a placebo in the trial, so all participants received a placebo. Patients allocated to hypericum also received a placebo that was indistinguishable from paroxetine (placebo paroxetine), whereas patients allocated to the paroxetine treatment group also received a placebo that was indistinguishable from hypericum (placebo hypericum). Patients did not know that in the placebo run in period they would receive only placebo and not one of the active treatments (hypericum or paroxetine) (c is true). To ensure that patients were blind to the allocation of placebo only, it was necessary to administer placebo hypericum and placebo paroxetine.

The placebo run in period allowed the researchers to ascertain which participants would be most likely to comply with treatment, thereby avoiding unnecessary randomisation of patients to treatment groups. In particular, some patients revoked their initial informed consent whereas others were lost to follow-up during the placebo run in period. However, selecting patients who were more likely to comply with treatment compromised external validity—the extent to which the sample represented the population. Nonetheless, selecting those patients who were more likely to comply with treatment may have promoted internal validity—the extent to which differences between treatment groups in outcomes can be ascribed to differences in treatment rather than differences in group characteristics. External validity and internal validity have been described in a previous question.3 Participants were not allowed to take other psychotropic drugs or receive psychotherapy during the study. The placebo run in period allowed some washing out of previous treatment. Presumably, the length of the washout period—between three and seven days—was the longest it was considered ethical for patients to have no treatment. In the example above, two patients were not allowed to proceed to the trial because they responded positively to placebo. If these patients had not been removed, it may have limited the generalisation of the study results because it would have been difficult to determine the extent to which outcomes were due to treatment rather than placebo.

Many trials are preceded by a single (patient) blind placebo run period during which all patients are given placebo. Such a design has been common practice within the drug industry and often recommended by standard texts. However, the placebo run in period has been criticised for being unethical. It has been described as deception, because patients think the trial has begun and are not aware that they will receive only placebo.


Cite this as: BMJ 2012;344:e3827


  • Competing interests: None declared.