Serial echocardiographic assessment of left ventricular mass:: how blinded should readers be?

Objectives In addition to the interest of mixing the sequence of echo-exam in a central blinded review, we studied the effect that might result from group-analysis of all echocardiograms simultaneously for each patient, with their sequence kept blind. A priori, this method of reading has the potential of decreasing measurement variability. Methods We included 630 echocardiograms from 210 hypertensive patients participating in a randomized clinical trial comparing two anti hypertensive agents for regression of left ventricular (LV) hypertrophy. Three echocardiograms per patient [selection (4 weeks before; W-4), at inclusion (week 0; W0), and the end of treatment (week 52; W52)], were read twice, according to two methods, blind to centre, patient numbers and sequence of visits: (1) examination of individual serial echocardiograms, (2) examination of all-patient mixed echocardiograms. The first method was expected to increase the power of treatment comparison by reducing variability of measurements of left ventricular mass (LVM). Results Pooling echocardiograms of all patients reduces variability of LVM change under treatment: absolute LVM (W52 - W0) standard deviation was reduced by 22%. Nevertheless, despite a good between-methods agreement for LVM values at each visit (intra-class coefficient of correlation from 0.88 to 0.92), LVM change under treatment was reduced even more, by 41%. Thus, the slight decrease of variability induced by gathering the echocardiograms is associated with an even greater reduction of LVM change. Conclusions According to these findings, the 'full-blind' methodology for a central blinded review in clinical trials appears to produce the maximum power of the study with the lowest sample size. (C) 2004 Lippincott Williams Wilkins.