Purpose: To perform a detailed analysis of blinding efficacy in the Investigational Vertebroplasty Efficacy and Safety Trial (INVEST) to determine any factors associated with patient unblinding.
Materials and methods: One hundred thirty-one patients were enrolled in this HIPAA-compliant, institutional review board-approved trial and were randomized to vertebroplasty (n = 68) or control intervention (n = 63). At multiple times up to 1 month, patients guessed their treatment allocation (vertebroplasty or control) with a forced-choice response. Patients offered a confidence level (scale, 0-10) for each guess and, when possible, a reason for their guess. Univariable logistic regression models were used to test for an association between baseline characteristics of the patients and correctly guessing their randomized treatment assignment. A two-sample t test was used to determine whether change in pain score differed between patients who guessed their treatment was vertebroplasty versus those who guessed their treatment was control intervention. Qualitative reasons for guesses were tabulated according to treatment assignment and guess.
Results: Recruitment site was the only baseline factor significantly associated with guessing treatment correctly at 14 days (P < .01). Between the control and vertebroplasty groups, a total of 54 patients offered 64 reasons for their guesses, with 16 (25%) reasons based on procedural experiences. Control patients guessing vertebroplasty had significantly greater pain improvement at days 14 and 30 than did those guessing control (day 14, P = .02; day 30, P < .001). In the vertebroplasty group, no relationship between change in pain and patient guess was noted.
Conclusion: A number of factors were associated with the ability of patients to guess their treatment allocation correctly-namely, treatment effect (in the control group only), study treatment site, and baseline pain duration. Procedural experiences rarely affected perceived treatment in INVEST.