Evaluation der Verblindung im Rahmen einer prospektiven randomisierten Studie mit dem Immunmodulator G-CSF bei Hochrisiko-Patienten (ASA 3 und 4) mit einem kolorektalen Karzinom

Abstract

Introduction: Blinding of doctors and patients in randomised clinical trials is an essential goal to reduce bias induced by expectations. Assessment of blinding is part of the CONSORT statement check-list for reporting of randomised trials [1]. We assessed blinding of clinicians and patients in a randomised controlled trial with G-CSF versus placebo to improve the postoperative outcome of high risk colorectal cancer patients [2]. Methods: A prospective, randomised, double-blinded trial was performed to test the effectiveness of the immune modulator G-CSF (filgrastim) to improve the postoperative outcome in 80 high risk patients (ASA 3 and 4) with colorectal cancer surgery. Study drugs (G-CSF or placebo) were injected s. c. 12 h before operation, 12 and 36 h after operation. To secure blinding, syringes were filled with the study drug (filgrastim or placebo) by ward nurses and not by the study surgeon, since shaking of ampulla’s with filgrastim would induce the production of some foam. Operating surgeons, house officers, study clinicians and patients were asked to guess in a test at 3 days after operation and at discharge what patients got — placebo or G-CSF. The size of the selection bias was assessed with phi-coefficients. Results: From the randomised 80 patients 75 were included in the analysis. At day 3 after operation and at hospital discharge similar results were obtained. Clinicians correctly rated in about 70% which study drug was applied (¤ Table 1). Patients were blinded, most (68/75) expected that they got G-CSF. Conclusions: Nearly all randomised controlled clinical trials using G-CSF as study drug claimed that they performed a double-blinded clinical trial [3], but there is no trial assessing blinding. Unblinding by surgeon and study observers was probably performed by knowledge of the patient cart including the leukocyte count. Patients were blinded without any problem. Positive expectation of the study patients can by one reason for the better recovery of these patients compared to patients not included into the trial [4]. Assessment of outcome parameters should be performed by a clinician not involved in the routine care of the patients, to guarantee an unbiased data recording and assessment.