Triple-Blind Study

Example 1

The first example, from a study by Abraham Heller and colleagues, highlights the difficulties in conducting a triple-blind study in a clinical setting of psychiatric patients (n = 15). The authors report using a triple-blind design to compare the effectiveness of three medications, which include imipramine, desipramine, and placebo. Participants were accepted into the study if the examiner determined that the individual was severely depressed, after which the examiner rated the individual on the Hamilton Depression Scale, and the participant completed self-reports on the Zung Self-Rating of Depression and the Minnesota Multiphasic Personality Inventory (MMPI). These scales were repeated in 1 week, biweekly during the hospitalization, and monthly for 3 months on an outpatient basis. The authors reported that a random triple-blind method was used to assign [Page 1542]medication to patients. Reportedly, a secretary was given the responsibility of pairing each participant's name to a list of three randomly assigned letters. After randomization, participants were monitored first in an inpatient treatment psychiatric setting, followed by treatment in an outpatient day-care setting for 1 – 4 weeks, and finally treatment in a regular outpatient setting. After a 3-month follow-up period, the patient was administered another MMPI and the previous scales. The authors stated that to eliminate bias no one with direct contact with the patients knew which medication the patient took during the study.

Example 2

A second example, from a study by Seung-Jung Park and colleagues, highlights the challenges in conducting a triple-blind study using multiple clinical sites. The study seems to be a much more rigorously conducted triple-blind study compared with the Heller study discussed earlier, given the methods described to ensure a triple-blind. The study examined in symptomatic patients (n = 177) whether a coronary stent coated with an antiproliferative agent paclitaxel prevented restenosis (e.g., whether this agent prevented hyperplasia and reoccurrences of coronary artery occlusion). The study involved three medical centers and compared stents coated with one of two doses of paclitaxel compared to a stent that was uncoated (i.e., placebo). Randomization was conducted using blocks of participants (i.e., each block used a 1:1:1 ratio to assign participants to 1 of the 3 interventions), and each block of participants was stratified according to stent diameter. Reportedly, stents were ordered in a randomized sequence, angiography was used to determine vessel diameter, and the appropriate diameter stent was selected. The authors state that patients, investigators, and core-analysis laboratories were all unaware to what groups the patients were assigned.

...