Adverse events during placebo vs. no drug administration—results of a randomised interventional trial in 160 volunteers

Dear Editor, A comparison to placebo in clinical trials is considered as gold standard to properly assess efficacy and tolerability of a drug if there is no established treatment for a druggable medical condition. However, it appears that the risk of patients related to the missed best proven treatment when randomised to placebo has often been underestimated [1]. Additionally, the nocebo risk, i.e. the risk that adverse effects may be also experienced by the mere expectation of receiving a drug, needs to be considered. Recent reviews on nocebo effects highlighted the paucity of conclusive experimental data in this area [2, 3]. This prompted us to re-evaluate the data of a relatively large population presented more narratively some time ago in two theses [4, 5], also considering the ethical dilemma related to acquisition of new data in this area. Two prospective randomised trials were conducted in 80 hospitalized and 80 non-hospitalized volunteers each. The volunteers provided written informed consent. Both groups consisted of 40 younger ( 60 years) individuals. Any current therapy was not influenced by the study. Study personnel wore white coats. During a run-in period of 6 days, every other day, all participants filled in standardized questionnaires (i) asking for the presence (yes/ no) of 20 common adverse symptoms during the last 2 days, with the possibility to enter additional symptoms [6] and (ii) quantifying actual anxiousness [7]. Further assessments included habitual anxiousness, introversion/extraversion, cognitive performance and general contentedness. After the run-in period, the volunteers were randomly allocated to a further observational-only period or to additional daily intake of a placebo tablet for further 6 days (Table 1). Volunteers in the placebo group were misleadingly informed that the study was conducted to assess the effect of a marketed preparation used for Bblood purification^ on potassium blood concentrations; accordingly, for complete camouflage, two blood samples were taken. Upon request for more information on the drug, it was explained that the preparation would not require an adjustment of current medication and that the effects would be achieved by mild diuresis. Joint statistical evaluation of all data was done using a general linear model, with type of intervention (placebo or none), hospitalization status, age group, sex (all fixed factors), age, the number of present medications and personality traits (all covariates) as independent variables and the individual changes in average (i) number of symptoms or (ii) results of actual anxiousness scales observed in the first (run-in) period and the second (intervention) period as dependent variables (IBM SPSS Statistics, version 20, IBM Corporation). As there was no prior estimation of the sample * Fritz Sorgel fritz.soergel@ibmp.net