Is the "end-of-study guess" a valid measure of sham blinding during transcranial direct current stimulation?

Studies using transcranial direct current stimulation (tDCS) typically incorporate a fade-in, short-stimulation, fade-out sham (placebo) protocol, which is assumed to be indistinct from a 10-30 min active protocol on the scalp. However, many studies report that participants can dissociate active stimulation from sham, even during low-intensity 1 mA currents. We recently identified differences in the perception of an active (10 min of 1 mA) and a sham (20 s of 1 mA) protocol that lasted for 5 min after the cessation of sham. In the present study we assessed whether delivery of a higher-intensity 2 mA current would exacerbate these differences. Two protocols were delivered to 32 adults in a double-blinded, within-subjects design (active: 10 min of 2 mA, and sham: 20 s of 2 mA), with the anode over the left primary motor cortex and the cathode on the right forehead. Participants were asked "Is the stimulation on?" and "How sure are you?" at 30 s intervals during and after stimulation. The differences between active and sham were more consistent and sustained during 2 mA than during 1 mA. We then quantified how well participants were able to track the presence and absence of stimulation (i.e. their sensitivity) during the experiment using cross-correlations. Current strength was a good classifier of sensitivity during active tDCS, but exhibited only moderate specificity during sham. The accuracy of the end-of-study guess was no better than chance at predicting sensitivity. Our results indicate that the traditional end-of-study guess poorly reflects the sensitivity of participants to stimulation, and may not be a valid method of assessing sham blinding.