Fd. Rubens et al., Feasibility of blinding in a randomized controlled trial comparing preoperative autologous blood donation and acute normovolemic hemodilution in adult cardiac surgery, TRANSFUSION, 40(9), 2000, pp. 1058-1062
BACKGROUND: Acute normovolemic hemodilution and preoperative autologous don
ation have been shown to be effective techniques for decreasing the exposur
e of patients to allogeneic blood during cardiac surgery. They have not, ho
wever, been compared to each other, because of perceived difficulties in bl
inding in such a clinical study. The feasibility of blinding was tested in
a pilot trial.
STUDY DESIGN AND METHODS: Ten patients were randomly assigned to undergo pr
eoperative autologous blood donation or acute normovolemic hemodilution dur
ing cardiac surgery. Patients were blinded during this process by shielding
of the arm and by insertion of an intravenous line in each patient. Every
attempt was made to blind the clinical staff during and after surgery. The
effectiveness of this blinding was determined by using a questionnaire.
RESULTS: In the 10 cases, six patients, four surgeons, and one anesthetist
answered, "I do not know," with respect to whether preoperative autologous
blood donation had occurred. The remaining people interviewed believed the
blinding was unsuccessful. However, correct answers were given by 75 percen
t of the patients (95% Cl, 19-99%), 83 percent of the surgeons (95% Cl, 36-
99.6%), and 66 percent of the anesthetists (95% Cl, 29.9-92.5%). The freque
ncy of correct answers did not differ significantly from the 50 percent exp
ected by chance, but the Cls are wide.
CONCLUSIONS: Blinding of patients and all members of the surgical team duri
ng both the preoperative donation process and acute normovolemic hemodiluti
on in the operating theater was successful most of the time, as the frequen
cy of correct answers did not differ significantly from the 50 percent expe
cted by chance. However, more accurate estimates of the success of blinding
require a study with a larger sample. Ii: is possible that, with a larger
series, the physician's ability to determine patient assignment would be si
gnificantly better than that by chance alone.